Who this ranking is for. Mid-market companies that will not run two AI stacks forever and need one primary harness usable by business users and engineers. For an engineering-weighted list of IDE/CLI/cloud coding harnesses, see the companion Top 10 harnesses when vibe coding gets unwieldy. That piece and this one share the same Security levels and alias scores (identical shared security levels in both articles).
Most companies will not want two AI stacks forever. One seat plan for sales and ops. Another for engineering. Different admins, different retention rules, different answers to “what left the building.” The frankenstack problem you already fight in RevOps shows up again as AI sprawl.
This ranking asks a different question than our coding harness list. There we score IDE agents and cloud coding tools for engineering teams (Top 10 harnesses when vibe coding gets unwieldy). Here we score platforms that can serve business users and engineers on equal footing: 50% of the rubric weight for business-user needs, 50% for engineering needs. Security (including prompt-injection controls) and governance sit in the shared half and are weighted hard on purpose.
Prices, ratings, and limits change. Every figure below was checked on October 30, 2026. Missing published data is marked not published. We never invent numbers.
TL;DR
Microsoft 365 Copilot + Microsoft Copilot Studio and GitHub Copilot Business/Enterprise tie at 82.8 (M365 takes #1 on business connectors; this is the office Copilot + Studio path, not GitHub Copilot and not Foundry Agent Service). Glean is third at 81.2. Cursor + Grok Bot is fourth at 78.8 after we scored the full Teams solution (Grok Bot desktop/mobile and Connect apps), not the IDE alone. ChatGPT Business/Enterprise + Codex ties that total at 78.8 and ranks fifth on the Reviews tie-break. Claude Team/Enterprise + Claude Code is sixth at 76.8. Google Gemini Enterprise is seventh at 76.0. Salesforce Agentforce is eighth at 70.8. Writer is ninth at 57.2.
Edition rule: every Security, Multi-model, and Admin evidence row must sit on the Business/Team seat we price. Admin 5 requires priced SSO + SCIM + audit logs + retention/spend controls. Scores assume a mid-market buyer on Business/Team editions; Enterprise editions can score higher on security and admin. See also the engineering-weighted coding harness ranking.
How we scored
Edition rule (editor decision, 2026-10-30). Score every tool on the same edition we price: the Business / Team / mid-market seat a mid-market company would buy, for all criteria including Security and Admin / identity / governance. Where Enterprise (or Microsoft 365 E5 / Purview Suite / Gemini Standard+) adds a level-5 deployment control (CMEK, EKM, customer VPC, dedicated single-tenant), mention it in prose as an upgrade path and do not score it. Applied to ChatGPT, Claude, Cursor + Grok Bot, Microsoft 365 Copilot, Gemini, Glean, GitHub Copilot, Agentforce, and Writer. Canonical Security levels are summarized in How we scored below (identical wording in the coding article). SSO and retention are recorded when sourced (or marked not published) and are not additional gates for levels 4-5.
Purview Customer Key. Microsoft Learn lists Microsoft 365 Copilot interactions under multi-workload Customer Key DEPs (overview). Licensing requires Microsoft 365 E5/A5/G5, Microsoft Purview Suite, or Office 365 E5 (Purview service description). That is not included with the Copilot $30/user/mo add-on alone, so Customer Key is an upgrade path and is not scored on the priced edition.
This is a product-harness ranking by design: we score commercial products mid-market teams can buy, not research agents alone. Vendor list economics apply (published seat and usage prices). We did not invent pass@k, pass^k, or dollars-per-task figures. OpenAI Verified findings and OWASP guidance are risk context, not a scored harness leaderboard. Scaffold-style pass-rate gaps are about 0 to 8 points (not rubric totals). Independent evals sometimes show cost spreads up to 40x across setups; we do not treat that as a per-tool score input.
Selection. Candidates had to be real in 2026, sell a team or enterprise surface, and plausibly cover both knowledge/business work and a coding or agent path for engineers. Included: Microsoft 365 Copilot / Copilot Studio, GitHub Copilot (Microsoft family; ownership cited), ChatGPT Business/Enterprise with Codex, Claude Team/Enterprise with Claude Code, Google Gemini Enterprise, Salesforce Agentforce, Cursor with Grok Bot, Glean, Writer, Amazon Q Business. Excluded from the scored set for this pass: pure vibe app builders, sunset coding SKUs already cut in the coding article, and vendor-only shortlists without a company-wide assistant story. Amazon Q Business is discussed outside the scored table with an availability caveat.
Rubric (weights sum to 100; revised during review).
| Category | Weight | Bucket |
|---|---|---|
| Price | 8 | Shared |
| Security (incl. prompt-injection controls) | 20 | Shared |
| Admin / identity / governance (SSO, SCIM, audit, retention) | 12 | Shared |
| Multi-model access | 8 | Shared |
| Business connectors (CRM, email, Slack/Teams, docs) | 12 | Business |
| Ease for non-technical users | 8 | Business |
| Coding / agent capability for engineers | 12 | Engineering |
| Engineering tooling (IDE/CLI/Git/CI/MCP) | 8 | Engineering |
| Team behind it | 6 | Shared |
| Real user reviews | 6 | Shared |
Business half = connectors + ease + half of shared = 50. Engineering half = coding + eng tooling + half of shared = 50. Security levels 4-5 require published prompt-injection or agent-tool controls (Model Armor, Prompt Shields, Trust Layer Prompt Defense, agent approvals, sandbox, MCP permissioning). Pure compliance without agent/PI = max Security 3. Level 5 needs a priced-edition deployment control (VPC/CMEK/EKM/dedicated) plus SOC 2 or ISO 27001, audit logs, and published PI/agent controls and VPC/CMEK/dedicated deployment; regional residency alone does not count. Full Security level definitions are summarized in How we scored (SSO and retention are recorded, not gates for Security 4-5).
External consensus. After excluding the Coworker vendor_editorial list (or treating it as non-primary), require n >= 3 where feasible; otherwise report n >= 2 without Coworker and flag thin coverage. Primary rank = equal weight per methodology class when multiple classes exist. Missing tools are skipped, not scored as last. See Consensus below and the consensus table.
Total-solution boundary. Score the commercially packaged product set a mid-market buyer gets under the named entry. Include Cursor + Grok Bot (Grok Bot on paid/Teams), ChatGPT + Codex + connectors, Claude Team/Enterprise + Claude Code + connectors, and Agentforce Slack-adjacent Salesforce collaboration. Keep GitHub Copilot alone. Score Microsoft 365 Copilot + Microsoft Copilot Studio without importing GitHub Copilot or Foundry. Score Gemini Enterprise without Gemini Enterprise Agent Platform (that build path is in the coding companion).
Microsoft Copilot Studio. First mention here: Microsoft Copilot Studio is the office Copilot agent builder path. It is not GitHub Copilot and not Foundry Agent Service.
Disclosure: coResolute uses Cursor and Grok Bot. We score every vendor with the same total-solution boundary, edition rule, and evidence standard. If Cursor + Grok Bot moves, it is because a surface included in the priced seat was previously scored as if it did not exist.
Data date. October 30, 2026 for most score cells; Cursor Biz/Ease rescore and method refresh November 2, 2026.
How we scored this, and how others do it
Coding task boards (SWE-bench, Terminal-Bench, Artificial Analysis, HAL) and security benchmarks (AgentDojo, InjecAgent) are widely cited in public “who’s best” lists. The OWASP Agentic Top 10 is a risk taxonomy, not an evaluation suite. Enterprise assistant mindshare lists (Rework, IT AI Index, G2 chatbots) favor front doors like ChatGPT and Microsoft 365 Copilot. Cost-aware agent research (Kapoor et al., arXiv 2407.01502; HAL) warns that accuracy-only ranks hide spend. Scaffold research (arXiv 2607.22585) shows harness choice can drive up to 40x token cost with pass-rate gaps of only 0-8 points.
This companion list is different on purpose: 50% business-user weight and 50% engineering weight. Security uses the same shared security rubric as the coding article (the same Security levels as the shared security rubric). We did not run SWE-bench, AgentDojo, or pass^k ourselves and we report no cost-per-task. Rubric wording was revised during review and logged in reconciliation.
Details and citations: Kapoor et al. (arXiv 2407.01502), Scaffold Effect (arXiv 2607.22585), OpenAI SWE-bench Verified notes, AgentDojo/InjecAgent, and the OWASP Agentic Top 10 (checked October 30, 2026).
Scorecard
| Rank | Tool | Price | Sec | Admin | Multi | Biz | Ease | Coding | Eng | Team | Reviews | Total |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Microsoft 365 Copilot + Microsoft Copilot Studio | 3 | 4 | 5 | 3 | 5 | 5 | 4 | 3 | 5 | 4 | 82.8 |
| 2 | GitHub Copilot Business/Enterprise | 5 | 4 | 5 | 5 | 2 | 2 | 5 | 5 | 5 | 4 | 82.8 |
| 3 | Glean | 2 | 5 | 5 | 4 | 5 | 4 | 3 | 2 | 4 | 5 | 81.2 |
| 4 | Cursor + Grok Bot (Teams/Enterprise) | 4 | 3 | 3 | 4 | 4 | 4 | 5 | 5 | 4 | 5 | 78.8 |
| 5 | ChatGPT Business/Enterprise + Codex | 4 | 3 | 3 | 4 | 4 | 5 | 5 | 4 | 5 | 4 | 78.8 |
| 6 | Claude Team/Enterprise + Claude Code | 4 | 3 | 3 | 2 | 4 | 4 | 5 | 5 | 5 | 5 | 76.8 |
| 7 | Google Gemini Enterprise | 4 | 4 | 3 | 4 | 5 | 4 | 3 | 3 | 5 | 3 | 76.0 |
| 8 | Salesforce Agentforce | 3 | 4 | 4 | 2 | 5 | 4 | 3 | 2 | 5 | 2 | 70.8 |
| 9 | Writer | 2 | 3 | 2 | 4 | 4 | 4 | 2 | 1 | 4 | 3 | 57.2 |
At equal 82.8, M365 beats GitHub on Business connectors (5 vs 2). At equal 78.8, Cursor ranks over ChatGPT on Reviews (5 vs 4). Glean 81.2 vs Gemini 76.0 follows Admin-rescored totals.
Consensus from other published rankings
Poll-aggregator style: average each tool’s rank across sources that name it.
Primary table (Coworker vendor_editorial excluded; n >= 3):
| Consensus rank | Tool | Unweighted avg | n sources |
|---|---|---|---|
| 1 | ChatGPT Enterprise | 1.33 | 3 |
| 2 | Google Gemini Enterprise | 4.00 | 3 |
| 3 | Claude Enterprise | 5.00 | 3 |
Sensitivity (n >= 2, still excluding Coworker): Microsoft 365 Copilot avg 1.50 (n=2), Glean 4.00 (n=2), Amazon Q Business 7.00 (n=2). With Coworker included at full weight, M365 averages 1.33 and ChatGPT 1.50, so M365 becomes #1 on that sensitivity table; we exclude Coworker from the primary average because it is vendor editorial for a competing assistant product.
Sources. Rework Best AI Tools for Enterprise (2026-07-16, editorial). IT AI Recommendation Index enterprise assistants (2026-10-04, share-based). G2 Learn best AI chatbots (2026-04-15, review_grid). Coworker (2026-10-15, vendor_editorial) recorded in the consensus table with include_in_primary=no. Method: unweighted mean among included sources; n >= 3 for the primary table; no last-place imputation.
Coverage note. Cursor is often missing from assistant-poll sources, so consensus will not confirm the Cursor rise; that is coverage, not a hidden rank.
Where consensus and this rubric disagree. Our primary consensus table (n>=3, Coworker excluded) puts ChatGPT first; this rubric ranks it fifth at 78.8 (tied with Cursor on total; loses the Reviews tie-break) after Admin is scored on the Business seat. Gemini holds Security 4 on the Business edition we price (Model Armor supports the PI gate; Standard/Plus CMEK is an upgrade path). Claude places sixth on the 50/50 score despite mid-pack assistant-poll ranks. Glean needs the n>=2 sensitivity table after Coworker is dropped. Amazon Q must not be a new-buy default after the July 31, 2026 freeze, and is out of our buyable top list entirely.
The ranked nine
The ranking
1. Microsoft 365 Copilot + Microsoft Copilot Studio (82.8)
What it is. Office Copilot + Microsoft Copilot Studio (not GitHub Copilot, not Foundry Agent Service). Microsoft’s add-on assistant inside Word, Excel, PowerPoint, Outlook, and Teams, with Copilot Studio / Agent Builder for custom agents. GitHub is owned by Microsoft (acquisition completed October 26, 2018; GitHub FAQ confirms ownership). GitHub Copilot is a separate SKU in the same family when you need IDE-depth coding.
Best for. Microsoft-standardized companies that need business users productive on day one, with Studio agents and an optional GitHub Copilot add for engineering.
Watch-outs. Enterprise add-on is $30 per user per month paid yearly and requires a qualifying Microsoft 365 license (microsoft.com Copilot enterprise pricing, checked 2026-10-30). Model choice is narrower than BYOK platforms. Treat “one Microsoft family” as a procurement story, not a single SKU.
Key facts (checked 2026-10-30). Enterprise Copilot add-on is $30/user/month (annual) on Microsoft’s Enterprise Copilot pricing page (microsoft.com/en-us/copilot/pricing/enterprise, checked October 30, 2026). The Business pricing page now describes Copilot bundled into Business Standard / Business Premium SKUs rather than a standalone Business add-on; we score the Enterprise add-on list path above and do not treat the Business bundle page as a separate seat price. Prompt defense in depth documented on Microsoft Learn; Defender offers real-time agent protection paths. G2 listing at microsoft-microsoft-365-copilot/reviews redirects to Microsoft Copilot and shows 4.3/5 from about 630 reviews as of October 30, 2026 (counts drift; G2 may 403).
Upgrade path. Purview Customer Key can encrypt Microsoft 365 Copilot interactions, but it requires M365 E5 / Purview Suite / O365 E5 licensing, not the Copilot $30 add-on alone. We do not score it on the priced edition.
Sourced checks (October 30, 2026). SOC 2; PI; audit; deployment (upgrade path; not scored); SSO.
2. GitHub Copilot Business/Enterprise (82.8)
What it is. Boundary = engineering SKU alone (Biz/Ease stay 2 by rule). GitHub’s coding assistant and agent stack. Microsoft-owned. Wins on engineering half; loses on business connectors and non-technical ease. Included so buyers see the pure engineering SKU next to M365 Copilot.
Best for. Engineering-led orgs already on GitHub Enterprise Cloud that will keep a separate business assistant, or that pair Copilot with M365 Copilot under Microsoft.
Key facts (checked 2026-10-30). Business $19 / Enterprise $39 per seat with pooled AI credits (coding article sources). Published cloud-agent prompt-injection mitigations (risks-and-mitigations). Data residency is regional routing, not dedicated VPC/CMEK deployment under our level-5 bar, so Security is 4 (aligned with the coding article). Multi-provider models plus BYOK with admin model/key controls on the Business path we price (Multi-model 5; BYOK). G2 4.4/5 from 400 reviews.
Sourced checks (October 30, 2026). SOC 2; PI; audit; multi admin / model availability; enable custom models; BYOK; SSO.
3. Glean (81.2)
What it is. Enterprise AI search, assistant, and agents grounded in a permissions-aware company graph across a large connector catalog.
Best for. Tool-sprawl companies where the first job is “answer from our systems” for every employee, with agents on top.
Watch-outs. Per-seat dollars are not published (Core Suite / Enterprise Flex structure is). Engineering tooling is not an IDE harness. Code Writer is a metered capability, not Claude Code or Codex.
Key facts (checked 2026-10-30). Published AI Security policies cover prompt injection; agent sandbox docs exist. G2 4.7/5 from 337 reviews (page meta dated 2026-10-27).
Priced edition. Security evidence uses the standard commercial Glean Protect deployment (contact-sales seats; Price 2 because seat dollars are not published). Docs describe single-tenant architecture across deployments (Glean Protect, checked October 30, 2026).
Sourced checks (October 30, 2026). SOC 2; PI; audit; multi-model; deployment.
4. Cursor + Grok Bot (Teams/Enterprise) (78.8)
What it is. Cursor IDE + Cloud Agents and Grok Bot (included on paid/Teams): desktop, iOS, Android, Connect apps, routines, optional Cloud Agent delegation. Cursor and Grok Bot are one entry under the total-solution rule because Grok Bot ships with paid individual plans and Cursor Teams (cursor.com/docs/grok-bot, cursor.com/help/grok-bot/connect-apps, cursor.com/docs/grok-bot/security).
Best for. Companies that want one vendor for an engineering harness plus a business-user agent without buying a second seat family. Still weaker than M365/Glean on org-wide permissions-aware connectors and Admin on Teams.
Watch-outs. Biz connectors 4 (member-OAuth Connect apps, not a Glean-class ACL graph). Ease 4 (own apps, not Outlook/Word chrome). Teams Admin still 3 (SCIM/audit Enterprise). Security still 3 on Teams. Enforce spend limits. Grok Bot Teams default-on with no product off switch (grok-bot/teams). Teams Premium is $30/user/mo on the primary Teams pricing page.
Admin (priced edition). Teams publishes SSO and chat retention; SCIM and audit logs are Enterprise-only (two gaps), so Admin is 3.
Upgrade path. CMEK, Enterprise audit logs, and Enterprise admin model controls are upgrade paths (not scored on Teams).
5. ChatGPT Business/Enterprise + Codex (78.8)
What it is. Tied with Cursor at 78.8; loses the Reviews tie-break (4 vs 5). Connectors remain in-boundary. OpenAI’s company workspace for chat, apps, company knowledge, and agents, with Codex as the coding agent under the same Business/Enterprise admin plane.
Best for. Organizations that want one familiar front door for business users and a real coding agent for engineers without splitting identity and spend.
Watch-outs. Enterprise seat dollars are not published (contact sales). Score Business list prices for self-serve teams, then negotiate Enterprise for SCIM, EKM (Enterprise upgrade path; not scored on Business), residency, and Compliance API. Connector coverage is strong but not as wide as Glean’s connector catalog.
Key facts (checked 2026-10-30). Business Standard $20 per user per month annual, $25 monthly (OpenAI business pricing). Codex included on Business and Enterprise paths used in our coding ranking. Agent approvals / security docs cover prompt-injection and agent-tool risks (developers.openai.com/codex/agent-approvals-security; redirects to ChatGPT Learn, content matches; checked October 30, 2026). SSO on Business; SCIM on Enterprise. Salesforce plugin and Microsoft 365 / Slack-style apps documented. G2 ChatGPT 4.6/5 from about 3,800 reviews as of October 30, 2026 (counts drift; G2 may 403); Codex product page smaller volume.
Admin (priced edition). Business publishes SAML/OIDC SSO; SCIM, Compliance API audit export, and custom retention are Enterprise upgrade paths, so Admin is 3 under the edition rule.
Upgrade path. EKM and Compliance API / deeper audit are Enterprise upgrade paths (not scored on Business).
Sourced checks (October 30, 2026). SOC 2; PI; audit (upgrade path; not scored); multi-model; deployment (upgrade path; not scored).
6. Claude Team/Enterprise + Claude Code (76.8)
What it is. Anthropic’s Claude Team/Enterprise seats covering Chat, Cowork, and Claude Code under one identity and spend controls.
Best for. Teams that want frontier Anthropic quality for documents and coding, plus business connectors (Microsoft 365, Slack, Salesforce, Google Workspace apps) and published prompt-injection guidance on Claude Code.
Watch-outs. Official docs: routing Claude Code to non-Claude models is unsupported (Multi-model 2). Usage bills at API rates on top of seats. Token burn is a real ops issue.
Key facts (checked 2026-10-30). Team Standard from $20/$25. Enterprise seat plus usage. Connectors documented in Claude Help Center. Security page includes prompt-injection controls. G2 Claude Code 4.7/5 from about 250 reviews as of October 30, 2026 (counts drift; G2 may 403).
Admin (priced edition). Team publishes SSO, domain capture, JIT, and spend controls; SCIM, audit logs, Compliance API, and custom retention are Enterprise upgrade paths, so Admin is 3.
Upgrade path. CMEK and Enterprise-cited SOC/audit/SSO packaging are upgrade paths (not scored on Team).
Sourced checks (October 30, 2026). SOC 2 (upgrade path; not scored); PI; audit (upgrade path; not scored); deployment (upgrade path; not scored); SSO on Team (priced; scored).
7. Google Gemini Enterprise (76.0)
Admin note. Business edition documents SSO-class Google identity and Cloud Audit Logs; SCIM (Workforce Identity Federation) and retention controls are not published on the Business security page we price, so Admin is 3 under the rubric bar (SSO + SCIM + audit + retention).
What it is. Gemini Enterprise front door (not Gemini Enterprise Agent Platform from the coding companion). Google’s intranet search, assistant, and agentic platform (Agentspace lineage), with edition-based connectors into Workspace and major SaaS systems.
Best for. Google Cloud / Workspace-standard orgs that want one Google front door for search, chat, and agents, with Model Armor as published PI (Security 4 on the Business SKU under the shared rubric).
Watch-outs. Gemini Code Assist Standard/Enterprise remains sunset for new buys; do not confuse SKUs. Prefer primary Google Cloud / Gemini Enterprise pricing pages over secondary comparisons before you sign.
Key facts (checked 2026-10-30). Gemini Enterprise Business edition starts at $21 per user per month on Google’s primary pages (gemini-enterprise/business, second source gemini-for-work). Model Armor screens prompts and responses and is documented for Gemini Enterprise (enable Model Armor); the Business edition page also names Model Armor under data security. Security is 4 for the Business edition we price ($21): Model Armor supplies PI, and SOC 2 plus audit rows exist, but CMEK is documented for Standard/Plus, not Business. Connectors cover Workspace, Microsoft 365, HubSpot, Slack, and more. Coding depth is thinner than Codex/Claude Code/GitHub Copilot on this 50/50 rubric.
Upgrade path. CMEK on Standard/Plus is an upgrade path relative to Business.
8. Salesforce Agentforce (70.8)
Admin note. Salesforce SSO and Generative AI / Trust Layer audit (Data 360 enabled) plus Setup Audit Trail 180-day retention are on the priced path; a live-readable Agentforce-specific SCIM page could not be verified, so Admin is 4 (one gap).
What it is. Agents inside Salesforce data and workflows, with Flex Credits and related pricing models, plus Agentforce Vibes for a coding-adjacent path.
Best for. Salesforce-centric RevOps and service orgs where write-back to CRM under existing object permissions is the point.
Watch-outs. Pricing models are easy to mis-model (conversations vs Flex Credits vs user licenses). Multi-model path is Salesforce/Einstein-centric (score 2). Product-specific G2 could not be verified cleanly this pass. Engineering IDE depth is not the product’s center of gravity.
Key facts (checked 2026-10-30). Einstein Trust Layer includes Prompt Defense for toxicity and prompt-injection detection (Salesforce Help: configure prompt injection detection). Security is 4 under the shared rubric (Trust Layer Prompt Defense / PI detection plus SOC 2 and audit; no customer VPC/CMEK/dedicated row; same Security 4 as the coding article). Prompt injection detection is called out as beta on related Salesforce Help pages. Pricing: see salesforce.com/agentforce/pricing.
Sourced checks (October 30, 2026). SOC 2; PI; audit; SSO.
9. Writer (57.2)
What it is. Enterprise agentic platform for governed content and workflows, with Knowledge Graph, agents, SSO/SCIM on Enterprise, multi-model catalog, and BYOM on Enterprise (Multi-model 4: BYOM without a sourced admin model-allowlist URL).
Best for. Marketing, communications, and content ops that need brand and claim control company-wide. Not a substitute for an engineering IDE harness.
Key facts (checked 2026-10-30). Starter lists features on writer.com/plans without a published seat dollar amount (checked October 30, 2026); Price is 2. BYOM via Bedrock/OpenRouter with documented pass-through. Security/trust pages cover SSO, SOC 2, guardrails; dedicated PI page not published this pass (Security 3).
Admin (priced edition). Starter (the priced mid-market seat) has no published SSO/SCIM/audit story; those controls are Enterprise, so Admin is 2.
Upgrade path. Enterprise SSO/SCIM and Enterprise audit controls are upgrade paths; Security 3 reflects missing published PI on the path we model.
Sourced checks (October 30, 2026). SOC 2; audit (upgrade path; not scored); multi-model; SSO (upgrade path; not scored).
Outside the buyable top list: Amazon Q Business (closed to new)
What it is. AWS’s business assistant with published Lite $3 and Pro $20 per-user pricing and SaaS connectors.
How to choose one
- Inventory the center of gravity. Microsoft 365, Google Workspace, Salesforce, or mixed SaaS sprawl.
- Decide whether engineers share the same front door. If yes, favor ChatGPT+Codex, Claude Team/Enterprise + Claude Code, or Microsoft family (M365 Copilot + GitHub Copilot). If engineers keep a dedicated IDE harness, say so explicitly and still pick one business front door.
- Demand published PI / agent controls before company-wide write actions.
- Model seat + usage, not seat alone.
- Run one pilot workflow that includes a business user and an engineer under the same admin plane.
For engineering-only depth rankings, use the coding companion. For CRM readiness before agents write to production objects, start with a free CRM Health Check.
FAQ
Who is #1? Microsoft 365 Copilot + Microsoft Copilot Studio at 82.8 (Security 4), ahead of GitHub Copilot on Business connectors.
Where did ChatGPT go? Fifth at 78.8 on the Reviews tie-break with Cursor (was fourth before the Cursor total-solution rescore). Admin stays 3 on Business-edition evidence.
Where did Cursor go? #4 at 78.8 after the total-solution rescore of Grok Bot connectors/ease (Biz 2→4, Ease 2→4). Security and Admin unchanged (both 3 on Teams).
Why did Cursor jump? Because the live score treated an included business agent as absent. Same rule includes ChatGPT and Claude connectors.
Amazon Q? Closed to new customers; omitted from the scored sheet (no synthetic rank). Discussed outside the scored table with an availability caveat.
Where is the engineering-weighted coding list? Top 10 harnesses when vibe coding gets unwieldy.
How current are prices? Checked October 30, 2026. Contact /contact/ if you want help piloting two options against your stack.
Sources
Checked October 30, 2026 unless noted.
- Coding companion: /top-10-harnesses-vibe-coding/
- Microsoft 365 Copilot enterprise pricing: https://www.microsoft.com/en-us/copilot/pricing/enterprise
- Microsoft 365 Copilot business pricing: https://www.microsoft.com/en-us/copilot/pricing/business
- Microsoft Copilot prompt defense in depth: https://learn.microsoft.com/en-us/microsoft-365/copilot/copilot-prompt-defense-in-depth
- Microsoft completes GitHub acquisition (2018-10-26): https://blogs.microsoft.com/blog/2018/10/26/microsoft-completes-github-acquisition/
- GitHub FAQ ownership: https://github.com/frequently-asked-questions
- OpenAI business pricing: https://openai.com/business/chatgpt-pricing/
- OpenAI Salesforce plugin: https://openai.com/business/plugins/salesforce/
- Claude Team plan: https://support.claude.com/en/articles/9266767-what-is-the-team-plan
- Claude connectors: https://support.claude.com/en/articles/11176164-use-connectors-to-extend-claude-s-capabilities
- Claude Salesforce org setup: https://support.claude.com/en/articles/16952184-set-up-salesforce-in-claude-for-your-organization
- Claude Code security: https://code.claude.com/docs/en/security
- Claude LLM gateway: https://code.claude.com/docs/en/llm-gateway
- Gemini Enterprise connectors: https://cloud.google.com/gemini-enterprise/connectors
- Glean Core Suite pricing structure: https://docs.glean.com/glean-core-suite-pricing
- Glean AI security: https://docs.glean.com/administration/protect/ai-security/introduction
- Glean agent sandbox: https://docs.glean.com/security/agent-sandbox-ptc
- G2 Glean: https://www.g2.com/products/glean-technologies-glean/reviews
- Writer plans: https://writer.com/plans/
- Writer AI Studio pricing / BYOM: https://dev.writer.com/home/pricing
- Salesforce Agentforce pricing: https://www.salesforce.com/agentforce/pricing/
- Amazon Q Business pricing: https://aws.amazon.com/q/business/pricing/
- Amazon Q Business availability change: https://docs.aws.amazon.com/amazonq/latest/qbusiness-ug/qbusiness-availability-change.html
- Cursor Teams pricing: https://cursor.com/docs/account/teams/pricing
- Cursor Grok Bot: https://cursor.com/docs/grok-bot
- Cursor Agent Security: https://cursor.com/docs/agent/security
- Rework enterprise AI tools 2026: https://resources.rework.com/tools/ai-tools/best-ai-tools-for-enterprise-2026
- IT AI Index enterprise assistants: https://it-ai-index.com/it-data/enterprise-ai-assistants/enterprise/
- Coworker enterprise AI assistant comparison: https://coworker.ai/blog/best-enterprise-ai-assistant-2026
- G2 Learn best AI chatbots: https://learn.g2.com/best-ai-chatbots
- iTechGuides enterprise assistants buyer guide (shortlist, not ranked): https://www.itechguides.com/best-enterprise-ai-assistants-for-organizations-with-thousands-of-employees/
- GitHub Copilot plans: https://docs.github.com/en/copilot/get-started/plans
- G2 ChatGPT: https://www.g2.com/products/chatgpt/reviews
- G2 Claude Code: https://www.g2.com/products/anthropic-claude-code/reviews
- G2 GitHub Copilot: https://www.g2.com/products/github-copilot/reviews
- G2 Microsoft 365 Copilot: https://www.g2.com/products/microsoft-microsoft-365-copilot/reviews
- G2 Cursor: https://www.g2.com/products/cursor/reviews
- https://learn.microsoft.com/en-us/purview/customer-key-overview
- https://docs.glean.com/administration/protect/overview
- https://developers.openai.com/codex/agent-approvals-security
- https://help.salesforce.com/s/articleView?language=en_US&id=ai.generative_ai_trust_configure_prompt_injection_detection.htm&type=5
