Genspark
by MainFunc, Inc.
An all-in-one AI workspace where autonomous agents hand back finished work, not chat replies
Genspark is an all-in-one AI workspace built around autonomous agents rather than a chatbot. Its Super Agent decomposes a goal into steps, selects among roughly 30 models and 80-plus integrated tools, executes them and returns a finished deliverable — a deck, a working site, a research brief or a completed phone call — for knowledge teams that want output instead of prompting.
Genspark is an all-in-one AI workspace built by MainFunc, Inc., a Palo Alto company founded in 2023 by Eric Jing (a former Baidu vice president and Microsoft Bing research director), Kay Zhu (ex-Google principal architect) and Wen Sang (an MIT PhD who previously founded and sold the SaaS company Smarking). It launched in 2024 as an AI search product and pivoted in April 2025 to Super Agent, an orchestration layer that decomposes a stated goal into steps, chooses among roughly 30 models and more than 80 integrated tools, executes them and returns a completed artifact rather than a chat transcript. The surface products built on that engine include AI Slides, which turns a prompt or outline into a formatted ten-slide deck in around a minute; AI Developer, which builds websites and applications and connects to GitHub so non-coders can ship functional pages; AI Meeting Notes for transcription and action items; image generation through FLUX and Imagen; video through Sora, Kling and Veo; and Call For Me, which places real outbound phone calls in a natural voice and holds a genuine back-and-forth. Commercially it has grown unusually fast — reported ARR crossed $250M within roughly twelve months of the Super Agent launch — and in June 2026 the company extended its Series B to $485M at a $2.6B post-money valuation, alongside a global partnership announced with Microsoft in April 2026 to embed Genspark agents into Word, Excel, PowerPoint and Agent 365. It competes directly with Manus and OpenAI's ChatGPT Agent, and its most contested attribute is commercial rather than technical: everything is metered in credits, and how that meter behaves is the dominant theme in public buyer reviews.
A knowledge-work team lead or ops manager who wants finished deliverables — decks, research briefs, working prototypes — produced end to end rather than a chat assistant staff have to steer prompt by prompt.
One subscription covers roughly 30 frontier models and 80-plus tools behind a single agent that executes multi-step work and returns the completed artifact.
At a Glance
- Category
- AI Agents & Orchestration
- Pricing
- Freemium, Subscription, Usage-based
- Target Market
- CTOs, CIOs, Enterprise Developers, Marketing Leaders, Operations Managers, Consultants
- Deployment
- Cloud-only
- Founded
- 2023
- Headquarters
- Palo Alto, United States
- Customers
- 1,000+ enterprise clients and 2M+ monthly active users (company-reported, June 2026)
Key Features
- ✓Super Agent
Autonomously plans a multi-step task, picks the right models and tools, executes them and returns the finished deliverable rather than instructions
- ✓Mixture-of-agents model routing
Routes each sub-task across roughly 30 models and 80-plus tools, so one subscription reaches frontier text, image and video models
- ✓AI Slides
Generates a formatted ten-slide presentation from a prompt or outline in about a minute, reusing existing decks as templates
- ✓AI Developer
Builds websites and applications from a description and connects to GitHub, letting non-engineers ship a functional page
- ✓Call For Me
Places real outbound phone calls with a natural voice and holds two-way conversation, reported at an 83% success rate on complex tasks
- ✓Team workspace with SSO/SAML
Self-serve Team plan adds SAML single sign-on, admin analytics and per-seat credit allocation for organisations of 2 to 150 users
Capabilities
Use Cases
- •Board and client deck production
A consultant converts a research outline into a formatted client deck in minutes instead of spending an afternoon in slide software
- •Competitive and market research briefs
An analyst issues one goal and receives a synthesised written brief with sourced findings, replacing a manual multi-tab search session
- •Rapid internal tool prototyping
An operations lead describes a small internal web app and gets working code pushed to GitHub without involving an engineering sprint
- •AI subscription consolidation
A small team replaces separate chat, image and video generation subscriptions with a single credit-metered workspace covering all three
- •Outbound phone task automation
An office manager delegates supplier availability calls and reservations to an agent that dials out and reports the outcome
Ideal For
Best For
- ✓Turning a rough outline or prompt into a formatted, presentation-ready slide deck in about a minute
- ✓Multi-step desk research that has to span web search, document reading and synthesis into one written brief
- ✓Non-engineers standing up a working website or small application, with GitHub connected for the code
- ✓Consolidating several separate AI subscriptions (chat, image, video, research) into a single seat
- ✓Outbound phone tasks such as reservations, availability checks and supplier inquiries via Call For Me
Not Ideal For
- ✗Teams that need predictable, auditable monthly spend — everything is metered in credits, failed and retried tasks are still charged, and unused credits do not roll over
- ✗Customer-support automation, where a purpose-built helpdesk agent with ticket routing, deflection reporting and guaranteed escalation paths is the right tool
- ✗Heavy, repetitive high-volume agent workloads, where credit burn escalates the effective seat cost far faster than a flat-rate competitor
- ✗Procurement teams that require published list pricing up front — consumer Plus and Pro dollar prices sit behind a sign-in wall
Deployment
Market & Ratings
1,000+ enterprise clients and 2M+ monthly active users (company-reported, June 2026)
Market Analysis
Pros
- ✓AI Slides is repeatedly singled out as best-in-class, producing a polished ten-slide deck in roughly a minute
- ✓One subscription reaches frontier text, image and video models, with premium chat and image generation credit-free through the end of 2026
- ✓Genuinely completes multi-step work end to end rather than returning instructions the user still has to execute
- ✓Team tier is clean and self-serve at a published $30/seat with SAML SSO and admin analytics
Cons
- ✗Credit burn is the dominant complaint in public reviews — users report exhausting a Plus allowance within days or even hours, with agent tasks draining it fastest
- ✗Failed and retried tasks still consume credits, and unused credits neither pool across a team nor roll over between months
- ✗Trustpilot sits at roughly 1.5 out of 5 across about 112 reviews, with around 84% one-star, clustering on opaque billing and slow or absent support
- ✗Consumer Plus and Pro dollar prices are hidden behind a sign-in wall, forcing buyers to rely on third-party blogs to budget
- ✗The zero-credit access to premium chat and image models expires 31 December 2026, leaving procurement without a known post-2026 cost
- ✗Practitioner scepticism on Hacker News about how much of the product is a thin orchestration layer over other vendors' models
Pricing
Free
$0
- ✓Roughly 100 credits per day
- ✓Access to Super Agent and core tools
Plus
From $24.99/mo
- ✓From 10,000 credits per month
- ✓Higher credit tiers scale the monthly price
- ✓Zero-credit access to top chat and image models through 31 Dec 2026
Pro
From $249.99/mo
- ✓From 125,000 credits per month
- ✓Priority access for heavy agent and video workloads
Team
From $30/seat/mo
- ✓2-150 seats, self-serve
- ✓12,000 credits and 60 GB storage per seat
- ✓SAML SSO and admin analytics
Enterprise
Contact for pricing
- ✓151+ seats
- ✓25,000 credits per seat
- ✓Custom data residency
- ✓99.9% uptime SLA
Everything is metered in credits and the sticker price is only half the cost picture. Plus starts around $24.99/month for 10,000 credits and Pro around $249.99/month for 125,000, but both are tiered — third-party trackers put Plus at roughly $25 to $75/month depending on the credit block — and the consumer dollar prices are gated behind a sign-in rather than published. Task cost varies enormously: video generation on premium models burns credits fastest, slides run 100-plus credits each, and AI calls consume roughly one credit per second, while chat and image generation are zero-credit only until 31 December 2026. Credits do not roll over, failed and retried tasks still draw down the allowance, and one-time consumer top-up packs were removed around March 2026, so running dry forces a permanent tier upgrade or a wait for the monthly reset. Team is the one transparently published tier at $30/seat/month, but seat credits do not pool across the team, which pushes admins toward add-on credit packs whose prices are not public.
Security & Compliance
Connect
Sources
This page was written from 6 sources, 6 on domains other than genspark.ai.
Stay Ahead of the Curve
Weekly enterprise AI insights for technology leaders. No spam, no vendor pitches—unsubscribe anytime.
SubscribeRelated Products
Sema4.ai
Enterprise AI agents for knowledge work, running natively in your Snowflake or cloud account
Sapiom
Agent infrastructure that routes, runs and meters AI agents in production
Itential FlowAI
Governed AI agents for network and infrastructure operations, with deterministic execution and full audit trails
OpenAI Presence
Deploy production-grade AI voice and chat agents with enterprise policies, guardrails and evals