Pomegra
Bolt.new bought San Francisco startup Dokai on Sept. 29, 2026, adding autonomous enterprise sales agents to its AI app builder. Terms were not disclosed.
Key Takeaways
- Bolt.new announced its first acquisition on Sept. 29, 2026: Dokai, a San Francisco builder of autonomous go-to-market agents.
- Price, structure and Dokai's headcount were not disclosed. Co-founder Gerry Fernando Patia and the team join Bolt's AI and machine learning group.
- The first enterprise features built on Dokai technology are due in late fall 2026.
Lead
Bolt.new, the AI app-building product from StackBlitz, announced the purchase of Dokai on Sept. 29, 2026. It is the company's first acquisition. Dokai built autonomous agents that carry out sales-side work inside enterprise systems: prospect research, account tiering and CRM enrichment. Bolt did not disclose a price, and no funding or valuation figures accompanied the deal.
What Did Bolt.new Actually Buy?
Bolt bought a team and a body of agent technology, and the structure points to a talent-plus-technology deal rather than a customer-led one. Patia and the Dokai staff move into Bolt's AI and ML organization, and Dokai's agent technology becomes part of the Bolt platform. Bolt has not described a continuing standalone Dokai product.
Dokai's agents took natural-language instructions and ran multi-step workflows across enterprise software without a person approving each step. According to the announcement, that production experience is what Bolt wanted. Dokai was backed by Character Capital and ERA, though the round sizes were not disclosed.
Why Would an App Builder Want Sales Agents?
Bolt wants generated applications to connect to the systems enterprises already run, such as CRMs and internal data sources, and to act inside them. A tool that produces a front end is easier to replace than one that can write into a company's customer records. Dokai gives Bolt a tested layer for that kind of write access.
The timing fits Bolt's growth. Bolt launched in 2024 and reached an estimated $40 million in annual recurring revenue by March 2025, roughly five months after launch. It raised a $105.5 million Series B in early 2025, led by Emergence Capital and GV, at a reported valuation near $700 million. StackBlitz put total funding at $135 million as of December 2025. Much of that user base is individual builders and small teams, which makes the enterprise push a bid to move up-market.
What Are the Risks for Enterprise Customers?
The main risk is agents holding broad permissions inside systems of record. Security guidance such as OWASP's LLM06:2025 classification of "excessive agency" describes the pattern: a model with too much functionality, access and autonomy can cause damage from a single bad output. The August 2025 incident involving compromised OAuth tokens from the Salesloft Drift integration, which exposed data from multiple Salesforce instances, shows how connected agents widen that exposure.
Bolt has not said which permission scopes the new features will request. It has also not said whether agent actions will appear in enterprise audit logs, or whether administrators can switch off autonomous actions per workspace. Existing Dokai customers received no published continuity statement, which leaves open what happens to integrations already live in their environments.
How Does This Compare With Other Agent Acqui-hires?
The deal follows a run of larger platforms buying small agent startups for engineering talent and specific capabilities. Meta's purchase of Stilla and Adobe's of Rilo are cited as precedents. The common pattern is a short announcement, an undisclosed price, a product roadmap measured in months and little detail on the acquired company's existing customers. Bolt's release follows that template.
For Bolt, the deal is a first test of whether it can integrate a company at all. An acquirer that has never done a deal before tends to keep the team small and the scope narrow, and the late fall target gives it roughly one quarter to show working software.
What Comes Next for Bolt's Enterprise Push?
Bolt expects the first Dokai-based enterprise capabilities in late fall 2026. The measure of success will be whether enterprise buyers accept agents with write access to customer data, and whether Bolt publishes governance controls alongside the features. Competing app builders and CRM vendors are adding their own agent layers, so the window for differentiation is short.
A larger signal would be a second acquisition. A repeat purchase would suggest Bolt is building a deliberate agent stack. Silence after late fall would suggest Dokai was a one-off hire of talent.
Outlook
Bolt.new's first acquisition is small in disclosed terms and large in stated intent: moving from generating apps to operating inside enterprise systems. Execution risk sits in permissions, auditability and the fate of Dokai's existing customers. The late fall release will show whether the technology transfers into a product enterprises will approve.
