How to Pitch Gradient
How Gradient, the AI-only seed fund founded with Google, leads pre-seed and seed rounds from its $220M Fund 5, the five themes it names, and what it does not publish.
Gradient at a glance
Checked against Gradient's own pages and filings, November 2026
- Firm
- Gradient, "The seed fund designed for founders in AI."1 Founded in 2017 "in close partnership with Google," which "remains a valued LP and long-term partner."2 San Francisco, per its SEC filing.3 Assets under management not published.
- Funds
- Fund 5: "officially announcing the launch of our oversubscribed $220 million Fund 5" (March 2026), the first to take outside limited partners.2 The amended SEC Form D for Gradient Fund V LP reports $227.2M sold as of June 16, 2026.3
- Stages
- Pre-seed and seed. "We lead investments at the earliest stages of innovation."1 The Fund 5 post describes the job as spending "real time with founders at pre-seed and seed" and being "the first institutional check."2
- First check
- Not published. No check size or ownership target appears on the site.
- Focus
- AI only: "focused exclusively on AI." Five themes in the Fund 5 post: B2B software "built AI-native from day one," AI agents, the AI developer stack, real-world AI "from biology to materials, from robotics to compute substrates," and "open source, small models, hybrid compute."2 The portfolio page: "Actively investing in every layer of AI."4
- Geography
- Not published. No region list or US-only rule appears on the site.
- Partners
- Darian Shirazi (general partner; "leading seed investments in category-defining AI companies"), Zach Bratun-Glennon (general partner; "AI / ML and data science, vertical software, B2B marketplaces, fintech, and more"), Andrew Brackin (partner; "vertical AI, healthcare, and the application layer"), and partners Clayton Petty, Denise Teng and Vig Sachidananda, whose profiles state no focus area. The team page lists no focus areas.5
- Recent seed leads
- Amounts are whole rounds, not Gradient's share.
- What it looks for
- "Technical rigor," "an obsession with quality," "a deep desire to build an iconic and durable business," and "the speed to iterate faster than the market can copy." It says it has "backed more than 500 AI founders across applied AI, infrastructure, and new category-defining businesses."2
- How to reach
- Not published. There is no pitch form, submission email or cold-inbound statement; the Fund 5 post ends with "Reach out or follow us on" X and LinkedIn.2 A Founder Portal link on the site requires a login.
- Process
- Not published.
- Gradient: Home
- Gradient: Announcing Fund 5 (March 2026)
- SEC Form D/A: Gradient Fund V LP (June 2026)
- Gradient: Portfolio
- Gradient: Team
- Gradient: Investing in Linkup (February 2026)
- Nango: Nango raises $7.5M led by Gradient (April 2026)
- Gradient: Nango, integration infrastructure for AI agents (March 2026)
- Gradient: Investing in Malachyte (August 2026)
- Gradient: Photon (October 2026)
- Gradient: Investing in Zeroset (October 2026)
Gradient is the AI-only seed fund that Google helped start in 2017, when, in the firm's own words, "artificial intelligence was more of a buzzword than a category." Its portfolio page lists Flatfile, FlutterFlow, Infisical, Chili Piper, Krea, CentML and Airspace Intelligence. In March 2026 it announced an oversubscribed $220 million fifth fund, the first to bring in limited partners beyond Google.
What Gradient Is
Gradient describes itself as "the seed fund designed for founders in AI" and says it leads "at the earliest stages of innovation." The Fund 5 post is the clearest statement of the model: spend real time with founders at pre-seed and seed, be the first institutional check, and stay focused exclusively on AI. Its 2026 record matches that: it led Linkup's $10 million seed and Nango's $7.5 million seed, and co-led Malachyte with Bessemer, Photon with A* and Zeroset's pre-seed with 2048 Ventures.
What the firm does not publish matters as much when you are deciding whether to approach it:
- No check size or ownership target
- No geographic rule
- No pitch form, submission email or stated inbound policy
- No decision process or timeline
What Gradient Looks For
1. Technical rigor. The first of four founder traits the Fund 5 post names. The firm's partners come from Google and from AI research and engineering, and the post frames the fund around founders who can build at the frontier.
2. An obsession with quality. The second trait, stated as plainly as that.
3. A durable business, not a demo. "A deep desire to build an iconic and durable business." Gradient's homepage puts it as championing "disruptive ideas that have the potential to redefine industries."
4. Speed. "The speed to iterate faster than the market can copy." In a category where a better foundation model can arrive any quarter, this is the trait the firm says it weighs.
5. One of five themes. B2B software built AI-native from day one; AI agents reshaping how work gets done; the AI developer stack; real-world AI from biology and materials to robotics and compute; and open source, small models and hybrid compute. The portfolio page says it invests in "every layer of AI," so the themes are a map, not a fence.
What Gradient Needs to See
Gradient publishes no traction thresholds. Its 2026 leads are a better guide than any rule: Linkup (search infrastructure for AI), Nango (integration infrastructure for agents), Photon and Zeroset were all developer-facing or infrastructure companies at seed or pre-seed, which is the "AI developer stack" and "AI agents" themes in practice. Malachyte, which brought recommendation systems from Spotify to e-commerce, is the applied layer. If you are building for the people who build AI, you are in the middle of the fund's own examples.
How to Get a Gradient Meeting
- The partner whose focus matches: the team page lists titles only, but three profiles state a focus. Darian Shirazi leads seed investments in category-defining AI companies; Zach Bratun-Glennon covers AI and ML, data science, vertical software, B2B marketplaces and fintech; Andrew Brackin covers vertical AI, healthcare and the application layer. The firm's blog posts are signed, so the partner who wrote about a company like yours is the one to find
- Portfolio founders: with more than 500 AI founders backed, a founder in the portfolio is the most likely path to a partner
- Co-investors: Gradient's 2026 rounds were shared with Bessemer, A*, 2048 Ventures, Accel, Quiet Capital and Haystack, among others
- Social: the Fund 5 post's only call to action is to reach out or follow on X and LinkedIn. That is the published front door, such as it is
What Gets You Passed
- Not AI: the firm says it is "focused exclusively on AI"; a company where AI is a feature rather than the product is outside its stated scope
- Past seed: Gradient's stated job is the first institutional check at pre-seed and seed
- Slow to ship: the firm names speed of iteration as a founder trait it looks for, and its themes are categories where competitors copy quickly
A second track: Pitch Protocol
Warm paths into a fund like Gradient take time. Pitch Protocol is not a route into Gradient; it is a second track you can run alongside your outreach. Applying is free, your own AI agent submits one application in about ten minutes with no web form, and the funds in our network whose thesis matches reach out to you directly. Apply to the First 100 Founders Cohort →
Founder FAQ
Which Gradient partner should I pitch for my sector?
Gradient's team page lists titles only; three profiles state a focus. Darian Shirazi (general partner): "leading seed investments in category-defining AI companies." Zach Bratun-Glennon (general partner): "AI / ML and data science, vertical software, B2B marketplaces, fintech, and more." Andrew Brackin (partner): "vertical AI, healthcare, and the application layer." Partners Clayton Petty, Denise Teng and Vig Sachidananda publish no focus area.
Does Gradient lead seed rounds?
Yes. Its homepage says "We lead investments at the earliest stages of innovation," and in 2026 it led Linkup's $10M seed and Nango's seed, and co-led Photon's seed and Zeroset's pre-seed. It also joins rounds led by others (Sapiom's seed was led by Accel).
What is Gradient's typical check size?
Gradient does not publish a check size or ownership target. The closest statement is in its Fund 5 post: it wants to "be the first institutional check."
Does Gradient invest outside the United States?
Gradient publishes no geographic rule; nothing on its homepage, Fund 5 post or portfolio page names regions or countries. It is based in San Francisco per its SEC filing.
How long does Gradient take to decide?
Gradient does not publish a decision process or timeline.
Does Gradient take cold inbound?
Gradient publishes no pitch form, submission email or statement about cold inbound. Its Fund 5 post ends with "Reach out or follow us on" X and LinkedIn, and the site's Founder Portal requires a login.
Does Gradient only invest in AI?
Yes. Its Fund 5 post says the firm is "focused exclusively on AI," its homepage calls it "The seed fund designed for founders in AI," and its portfolio page says it is "Actively investing in every layer of AI." Five themes are named: AI-native B2B software, AI agents, the AI developer stack, real-world AI, and open source, small models and hybrid compute.
Frequently Asked Questions
Is Gradient the same as Google Ventures?
No. Gradient was founded in 2017 "in close partnership with Google" and says Google "remains a valued LP and long-term partner" of its $220M Fund 5, the first with outside limited partners. GV is Alphabet's own venture arm with a separate team and fund; see our GV guide.
Can I apply through Pitch Protocol instead of waiting for a warm intro?
Pitch Protocol is not a back door into Gradient, and we do not share which funds review an application before it is in. What it does: applying is free, your own AI agent (Claude, ChatGPT, Cursor or any agent that can read a URL) submits one application in about ten minutes with no web form, and the funds in our network whose thesis matches reach out to you directly. Founders anywhere can apply. Keep working your warm path to Gradient in parallel. Apply to the First 100 Founders Cohort →