Deepsky
San Francisco, US Β· 2 known investors
Acquired by Airtable October 2025 Β· terms undisclosed β Superagent technology and team for standalone AI research and analysis. Β· source β
DeepSky provides AI agents that handle research, drafting, content creation, and code development tasks for teams. The platform integrates with existing team workflows and tools, delivering live applications and artifacts that update as business needs change.
Also known as DeepSky Β· Hyperagent
Founders & leadership
Investors Β· 2
Company profile
researched Aug 2026DeepSky operates an agentic AI platform, marketed as Hyperagent, in which users assign a brief and a fleet of agents researches, drafts, builds and delivers the resulting work. Each agent runs in its own computing environment with access to real browsers, shells and connections to a customer's existing stack, and the platform exposes its execution steps β searches, decisions and artifacts β as they occur, while accumulating memories and skills over time. Outputs span multiple formats, including fully functional hosted websites and HTML pages, video, slide decks, documents and dashboards, with example use cases such as out-of-home advertising campaign planning, real estate marketing videos, candidate sourcing and ranking dashboards, and brand identity and launch-site packages. Delivered work is designed to remain current rather than static, refreshing as underlying conditions change.
The company began with investment and market research as its initial domain. According to the founding team, DeepSky integrates directly with data sources including FactSet, Crunchbase, Polygon and SEC filings so that users can query market, financial and company data in real time within the system, with the agent ranking and filtering sources and structuring findings into narratives. The product offers a "quick chat" mode for fast fact-finding alongside a deeper reasoning mode for full reports. The founding team is described as comprising Christopher Chang, a former investment banker and fundamental investor, CTO Forrest Moret, and Chief Architect Mark Huang, who came from quantitative trading; Michael Unger serves as head of sales. The team states it was among the first to extend model context length to one million tokens with high retrievable accuracy.
A Hugging Face organization page under the handle gradientai is presented as DeepSky, describing the company as a general business agent for knowledge workers and hosting a collection of long-context text-generation models in the 8B and 71B parameter ranges, last updated in 2024.
Founding story
Christopher Chang left investment banking to co-found the company after he and his co-founders β CTO Forrest Moret and Chief Architect Mark Huang, who had a background in quantitative trading β observed that the late-2022 release of ChatGPT had rapidly changed corporate information work while available models performed poorly on finance. The team began tuning models for the finance sector and extending context length, concluding that with the right tools, data and reasoning ability such models could support agentic systems.
Business model
DeepSky sells access to its agent platform, including a paid "DeepSky Pro" tier valued at roughly $240 per user for 12 months, alongside consumption of AI usage credits. Individual agent-produced artifacts shown on the company site are associated with per-task run times and dollar costs ranging from about $3.88 to $24.79.
Subscription access to DeepSky Pro plus usage-based AI credits, with per-task compute cost surfaced against each delivered artifact.
Traction
Approximately 300 MBA users, including many at Stanford GSB, were reported as early adopters as of August 2025, with their feedback prompting the addition of a quick chat mode alongside the deeper reasoning mode.
Latest developments
In August 2025 the company detailed an MBA Strategist Program for business school campuses and described early product changes driven by student users, including the introduction of a quick chat mode.
βΈFull profile β market position, technology, go-to-market, geography
Market position
Positioned as a general business agent for knowledge work that starts from high-stakes financial and market research, which the company characterizes as the most demanding form of information work, and extends into broader business execution.
Emphasis on source ranking and filtering to reduce hallucination, citation-backed structured narratives, direct integration with licensed financial and company data sources, transparent step-by-step execution, dedicated compute environments per agent, and deliverables that are continuously refreshed rather than one-off outputs.
Technology
Agentic system in which each agent is provisioned with its own computing environment, real browser and shell access, and connectivity to customer tools and data. The platform integrates with FactSet, Crunchbase, Polygon and SEC filings for real-time market and company data, ranks and filters sources, and produces structured narratives, live-rendered HTML sites, video, slides, documents and dashboards. The team reports early work on extending context length to one million tokens with high retrievable accuracy, and a related Hugging Face organization publishes long-context 8B and 71B text-generation models. Execution is exposed step by step, and the system generates new memories and skills as it works.
Go-to-market
Distribution runs through the tools teams already use: agents can be @-mentioned in Slack, connected to Telegram, triggered by webhook or run on a schedule, with deliverables sent to Slack or email. The company also runs a campus motion, an MBA Strategist Program, offering entire classes 12 months of free DeepSky Pro, early access to unreleased agent features, $1,000 in AI usage credits per student, referral bonuses of up to $1,000, performance-based campus prizes including a funded meeting with technology and venture capital leaders at the company's Silicon Valley headquarters, mentorship from its GTM, product and engineering teams, and up to $1,500 in funding for student-hosted AI events.
Knowledge workers and teams performing research, analysis and content production, with early adoption concentrated among MBA students β roughly 300 users including many at Stanford GSB β using it for case interview preparation, industry learning and internship projects. The company also positions the platform for professionals such as fund managers, solo entrepreneurs and executives evaluating competitive positioning.
Geography
Headquarters in Silicon Valley, United States, with campus outreach directed at US and international business schools.
Compiled by commissioned research from 8 cited public sources β announcements, filings, and press listed under research sources below.
Key figures
latest reportedCompany-reported or press-reported figures, each dated to when it was claimed β not independently audited.
Timeline Β· 2
launches, deals, and filingsDeepSky began rolling out an MBA Strategist Program offering business school classes 12 months of free DeepSky Pro, $1,000 in AI usage credits per student, early access to unreleased agent features, referral bonuses, campus event funding of up to $1,500 and mentorship from the company's GTM, product and engineering teams.
Following feedback from early MBA users, DeepSky introduced a quick chat mode for fast fact-finding in addition to its deeper reasoning mode for full-scale reports.
Dated company events from announcements, filings, and press; legal rows summarize public dockets and regulator releases.
βΈResearch sources Β· 8
primary sources listed
- Deepskydeepsky.ai Β· web
8 public sources were cited for this profile; the first-party ones are listed here.
Frequently asked questions
- What does Deepsky do?
- DeepSky builds AI agent fleets that research, build and continuously update deliverables inside a team's existing tools.
- Who founded Deepsky?
- Deepsky was founded by Mark Kim-Huang.
- Who are Deepsky's investors?
- Deepsky's investors include Icon Ventures, Mango Capital.
- Where is Deepsky headquartered?
- Deepsky is headquartered in San Francisco, US.