The partnership sells governed agentic AI
HCLTech — India's third-largest IT services firm — announced a Global Premier Partnership with Anthropic: a dedicated Anthropic Business Unit with a Center of Excellence to build agentic AI solutions for enterprises, embedding Claude, Cowork, and Claude Code into industry offerings and into HCLSoftware's own products (BigFix, AppScan), all under the banner of "governed Agentic AI."
HCLTech has planted a flag on our thesis and is selling enterprises the governed version. The market conversation has officially moved from "should we use agents?" to "who do we trust to build them right?" — which is the conversation a boutique agency wins.
But read the fine print, because the press release is selling the wrapper, not the work.
Framework governance misses the client's own systems
First: governance as a business unit. The partnership's machinery is a CoE, forward-deployed engineers, secure sandboxes, workforce training — governance delivered as a services offering. That machinery is genuinely useful, but it governs at the framework level: policies, guardrails, exception handling around AI Force workflows. The risk that actually kills enterprise agents lives one level down, in the client's specific systems — the proprietary database with no clean API, the clinical workflow no vendor product covers, the edge case your framework never modeled. Governance can't be shipped by a center of excellence; it has to be designed into the integration, system by system, by people who know the system.
Product-embedded agents skip the messy middle
Second: product-embedded agents solve the easy distribution. HCLSoftware embedding Claude into BigFix and AppScan means agents arrive inside products enterprises already buy — zero integration required, governance pre-installed. That's elegant, and it's also the least interesting part of enterprise AI. The workflows where agents would pay off most — the hospital's discharge process, the logistics team's exception handling, the internal tooling no ISV will ever cover — don't live inside a vendor's product. Product-shaped agents never reach the messy middle, and the messy middle is where the value is.
A business unit does not close the production gap
Third: the giant's admission, again. 54% of Indian organisations are deploying AI agents, but only 11% have moved to autonomous workflows (ServiceNow Enterprise AI Maturity Index 2026). HCLTech's announcement is another giant's answer to that gap — and it names the same cause: the tools are commoditized, the capability to make them work in production is not. A dedicated business unit doesn't fix unclear scoping, missing evals, or nobody owning the loop after launch. It just sells the fix at enterprise scale.
This is the gap our delivery model was built around. Giants sell governance as infrastructure and agents as products; someone still has to wire them into your actual systems, write the evals that prove they work, and stay accountable when they drift. That work doesn't scale down to a CoE template — it scales down to a team small enough to know your stack and stubborn enough to own outcomes.
HCLTech × Anthropic is good news for us, not bad. Every giant that endorses "governed agentic AI" moves more enterprise buyers from pilot-thinking to production-thinking — and production is where we live.
Field note: the discharge-summary agent
Our HMIS Pro work is the counterpoint: the kind of agent we design for HMIS Pro is meant for a regulated, high-stakes environment no product or platform touches — every clinical assertion linked to its source field, uncertain items rendered for review, the physician signing as author of record. No center of excellence ships with that. It is designed for the system, by people who know the system. That's the work giants can't productize — and it's the work we do at QuantaloomAI.
*Sources: HCLTech–Anthropic Global Premier Partnership announcement, Oct 12 2026 — Reuters, Morningstar/PR Newswire, The Hindu BusinessLine*
Related service Workflow automation
