If you run a Boost.ai virtual agent, its new owner has said nothing about your price, roadmap or data location, and you have until the Q4 2026 close to get answers in writing. Automation Anywhere announced a definitive agreement to acquire Boost.ai from Nordic Capital on October 7, 2026. The release names advisors and a closing quarter. It does not name a price, a product plan, a brand plan or a single commitment to existing customers.
The deal also gives Automation Anywhere its second conversational AI stack in under a year. If you bought Aisera, you should be asking which of the two survives.
What Automation Anywhere Actually Announced
The announcement is a strategy statement with a closing window, and very little else. Per the Automation Anywhere press release, the transaction is expected to close in Q4 2026, subject to regulatory approvals. BofA Securities advised Automation Anywhere, Santander advised Boost.ai, and Winston Taylor is Automation Anywhere's M&A counsel. Financial terms were not disclosed.
The pitch is plumbing. Boost.ai's chat and voice agents take the customer's request, and Automation Anywhere's automation platform executes the back-office work behind it. Boost.ai CEO Jerry Haywood framed it as connecting "the conversation, including voice, directly to the systems and people who get the work done." Automation Anywhere CEO Mihir Shukla said the combined company intends to give every enterprise "an autonomous operation that listens, decides, and acts."
What Boost.ai brings, by its own and the acquirer's description:
- "Hundreds" of customer organizations, strongest in Europe, concentrated in financial services, telecommunications and insurance (press release).
- Support for 36+ languages, claimed GDPR compliance, HIPAA alignment and an independent SOC 2 report.
- A 90%+ resolution rate in production, which is a vendor claim and should be read as one.
Unite.ai's write-up adds Automation Anywhere's own context: AI accounted for nearly 70% of its new and upsell bookings over the past six quarters, and agentic executions grew 5x year over year. That shows where the acquirer's sales force is pointed; the release says nothing about existing Boost.ai contracts.
Nordic Capital is the seller. It announced its partnership with Boost.ai on March 8, 2021, when Boost.ai had just over 100 employees, roughly NOK 100 million in ARR at the end of 2020, and named customers including Nordea, Telenor, Santander and DNB. A five-year private equity hold ending in a trade sale is ordinary. It also means the person who negotiated your current terms answers to a new owner within a quarter.
Who Is Exposed: Europe's Regulated Front Doors
The customers with the most at stake are the large European banks and insurers that put Boost.ai in front of millions of people. Nordea is the clearest public example. Boost.ai's Nordea case study describes 12 AI agents across Sweden, Denmark, Norway and Finland, more than 220,000 private banking conversations a month and a bank serving about 9 million private customers. The case study gives in-scope resolution rates of 91% for private banking and 95% for corporate banking in Sweden, figures Private Banker International also reported.
Boost.ai's own customer page lists Telenor, DNB, Ageas and Tryg alongside US names like MSU Federal Credit Union and Aspire General Insurance.
These are production deployments. A virtual agent handling a bank's first-line service is a production dependency with a regulator watching. If the integration roadmap moves hosting, swaps a model provider or pushes customers toward Automation Anywhere's cloud, each of those is a change your compliance team has to approve, and none of it is described in the announcement.
Why Aisera Customers Should Ask Which Stack Survives
Automation Anywhere already owns a conversational agent product, and the two overlap. It acquired Aisera on November 4, 2025, a provider of self-service AI agents for IT service management, HR and customer service, also on undisclosed terms. Today Automation Anywhere's Aisera page markets "Aisera for ITSM", "Aisera for IT Operations" and "Aisera for IT Service Desk", with no published customer FAQ on contracts, pricing or support transitions.
The obvious reading is a split: Aisera for employee service, Boost.ai for customer-facing chat and voice. That is plausible. It is also an inference, because neither announcement says so. Customer service sits on both product lists, and two dialogue engines, two intent models and two analytics layers inside one company rarely last.
If you are an Aisera customer using it for customer service, or a Boost.ai customer whose IT team was shown Aisera last year, ask your account team for the product boundary in writing, naming which product handles customer service and for how long. "Both products remain fully supported" does not answer that.
The Pricing Model Is the Real Exposure
Automation Anywhere has already said it wants to charge for AI agents by the work they do. When it bought Aisera, Shukla said "with AI agents doing the work, up to 40 percent fewer ITSM seats are needed" and that the company would charge for work performed instead of seat licenses, per DestinationCRM.
For a Boost.ai customer, that matters more than the logo on the invoice. Outcome or execution pricing can be cheaper, but it moves forecasting risk onto you: a seasonal spike in claims or card-fraud calls becomes a variable bill. Our buyer's guide to AI contact center platforms covers why "resolution" counts are the number vendors define most generously, and a per-resolution meter inherits that definition.
The fix is contractual. Hold your current price metric through the next renewal term, and make any switch to execution-based pricing opt-in, with your own definition of a billable resolution.
What GDPR Article 28 Already Gives You
European customers have a lever written into law. Under GDPR Article 28(2), a processor "shall not engage another processor without prior specific or general written authorisation of the controller," and where authorisation is general, the processor must inform the controller of any intended addition or replacement of other processors, "thereby giving the controller the opportunity to object."
A change in Boost.ai's owner does not by itself trigger that clause. Moving workloads onto Automation Anywhere infrastructure, adding its entities as sub-processors, or changing the language model behind the agent does. Read your DPA now for the notice period and the objection mechanics, so you can object inside the window when the notice arrives. We walked through the same mechanics when Stripe bought OpenRouter and when Adlib bought Paperbox, where DORA's contract rules sit on top for EU financial entities.
What Happened the Last Time a Conversational AI Leader Sold
The nearest precedent is NICE and Cognigy, and it ran on a similar clock. NICE announced the $955 million deal on July 28, 2025 and closed it on September 8, 2025; Cognigy became "NiCE Cognigy," a business unit folded toward NICE's CXone Mpower platform, with its co-founder running it.
That is the template to plan around. The product usually survives the first year under its own name. The commercial direction shifts toward the acquirer's platform, and the bundle becomes the default offer at renewal. If you run Cognigy or Kore.ai as a second source, keep that relationship warm through the close. If you run UiPath for back-office automation, expect Automation Anywhere to pitch Boost.ai as the reason to consolidate onto its stack. Weigh that pitch against how much switching both layers at once would cost you.
What to Do Before the Q4 Close
This Week:
- Pull every Boost.ai order form and DPA. Find the assignment and change-of-control clauses, the renewal date and the auto-renew notice window. Our guide to AI vendor exit clauses lists what standard terms do and do not give you.
- Ask your Boost.ai account manager three written questions: will the product, brand and support model be unchanged for 24 months; will hosting region and sub-processors stay as they are; and will current pricing hold through the next term.
This Month:
- Export your intent library, training data, conversation flows and integration configs, and confirm the format. A conversation design built over years is the asset you would lose in a forced migration.
- If you also own Aisera, map which use cases each product covers, and ask Automation Anywhere for the product boundary on customer service in writing.
Before Renewal:
- Negotiate a price hold on the current metric, an opt-in clause for any move to execution-based pricing, and a termination right if hosting or sub-processors change without your approval.
- For EU banks and insurers, run the deal through your ICT third-party register now, so the change is documented before the close. The HCL and Robotiq.ai piece covers the DORA angle for banks running acquired bots.
The Bottom Line
Automation Anywhere is building a front door for its automation platform, and Boost.ai's regulated European customers are the door. That strategy may be good for them: a virtual agent that can actually close a ticket in the back office is worth more than one that hands off to a human. But the release gives customers no commitments, and Aisera, a year in, still has no public customer FAQ. The usual sequence after a deal like this, seen with Inworld and Ultravox last week, is that terms get written by the acquirer at renewal unless the customer writes them first.
Send the three questions to your Boost.ai account team before the deal closes.
Continue Reading
- AI Vendor Exit Clauses: Standard Terms Give 0 to 90 Days to Get Out
- HCL Buys Robotiq.ai for €9M, Silent on the Banks Running Its Bots
- Best AI Contact Center Platforms: Containment Isn't Resolution
- Inworld Buys Ultravox and Upgrades Only Its Own Voices
- Stripe Bought OpenRouter. A Toggle Is Not a Contract.
- AI Agents Cut ITSM Costs 50% While Resolving 80% of Tickets