Obviously AI.

Obviously AI at a glance

Obviously AI was a San Francisco-based, privately held B2B software venture. Its original product let non-programmers train predictive models on structured business data. Its successor, Zams, focused on orchestrating AI agents across sales and revenue workflows. The original company’s exact legal name remained Obviously AI, Inc., trading as Zams in 2026. [4][5]

Evidence: Zams terms, founders' college announcement, original company page and 2025 rebrand statement. [4][6][7]

About Obviously AI

Short company description (55 words): Obviously AI was a no-code predictive analytics company that helped business users create machine-learning models from historical data without programming. Teams could upload or connect structured datasets, choose an outcome such as churn or sales, examine results, and generate predictions. The company later shifted toward automated AI agents under the Zams brand before announcing its wind-down in 2026.

Traditional BI primarily describes past performance through dashboards and metrics. Obviously AI sought to answer what is likely to happen next using supervised learning. That does not make its historical no-code models interchangeable with modern LLM-based agents.

Who founded Obviously AI?

Nirman Dave and Tapojit Debnath met at Hampshire College. Dave had worked on customer-behavior prediction at Streamlabs; Debnath had machine-learning experience at retail startup b8ta. Their combined experience shaped the original user-friendly AutoML product. Hampshire College says they started the venture in 2018. [6][9]

Leadership beyond the founders

The public sources checked do not support a complete current 2026 leadership roster. Historic CEO and CTO titles should not be presented as active appointments after the announced wind-down. A 2023 SEC filing identifies the two founders among company officers and Kiran Mysore Ravi as a director; a 2021 financing announcement says TMV took a board role. Later full board composition is not publicly confirmed. [10][11]

Founding story and company history

The founders saw a mismatch: business users were asking predictive questions, but existing tools were built for programmers. They developed the idea together at Hampshire, then worked in an Airbnb basement in the Bay Area as the company took shape. The startup passed through UC Berkeley SkyDeck before moving from beta toward a commercial product. [6][9]

Historical chronology: [6][10][9][12][13][1][2]

Products and features: past versus later

Self-service, managed service, or enterprise platform?

Answer: a hybrid model over time. Obviously AI combined no-code SaaS with optional hands-on data science; 2024–2025-era pricing pages offered software plus dedicated data-scientist and strategist support. Zams subsequently emphasized enterprise agent orchestration and sales-led deployment. It was not exclusively a simple spreadsheet-to-model app, nor solely a consulting firm. [14]

How Obviously AI built a predictive model

Historical company documentation describes a supervised machine-learning workflow, where historical records include the outcome to be learned. A user did not need to write Python, but still needed a meaningful target variable and correctly structured business data. [15][8]

Data requirements and input methods

The 2021 data-requisites page recommended one structured table or file, at least 1,000 rows and five columns, and an identifiable target/outcome column. Its CSV guide specifies a 25 MB file limit for that particular historical workflow; this is not a universal data cap across all later plans. Personally identifying information was not required. [15]

AutoML technology and algorithms

Obviously AI described its proprietary Edge Sharp AutoML pipeline as extracting dataset characteristics (including data types, dimensions, sparsity and variance), shortlisting five suitable algorithm families, testing hyperparameter configurations and choosing a model on reported accuracy/performance. Its publicly described model-search counts and speed are company-reported marketing claims, not independent benchmark results. [16]

Which algorithms were actually documented?

The official algorithms page confirms multiple machine-learning algorithms and tuning but presents the specific list within an image that could not be reliably transcribed for this review. It would be misleading to state that XGBoost, LightGBM, CatBoost, random forests, logistic regression or neural networks were each available without separately verifiable documentation. [16]

Explainability and performance

Official material described influence of underlying factors, shareable model reports and what-if simulations. These are useful explanations of correlation and model behavior, not proof of causal effect. Accuracy was discussed for classification; a complete historical checklist of F1, precision, recall, AUC, RMSE, MAE, R², confidence intervals and drift dashboards could not be verified across every plan. Always evaluate validation design and holdout performance before relying on a result. [10][17]

Prediction types and real-world uses

Classification, regression and time series are explicitly advertised on the original homepage. Individual applications are examples, not guaranteed product performance. [7][14]

Predictive machine learning versus generative AI agents

Zams officially positioned itself as an enterprise AI-agent orchestrator, not merely a refreshed version of a regression tool. Its 2026 legal terms document autonomous agents, task automation and knowledge assistants. Exact underlying OpenAI, Anthropic, Google or other foundation models were not established from the official public materials reviewed. Do not assign a specific LLM or model family without disclosure. [1][4]

API, integrations and deployment

The original site explicitly supported hosted shareable model web apps, real-time REST APIs and Zapier automation. The 2021 funding announcement referred to Adalo and Zapier integrations. The official homepage mentioned external presentation of predictions in BI tools such as Power BI and Looker; this does not establish native Power BI or Looker connectors. Developers should not rely on historical endpoints after the wind-down. [7][10]

Developer details: Public API authentication, limits, pricing, SDKs and batch endpoint specifications could not be checked because the historical developer portal was not accessible during this research. REST support is confirmed; a current usable API is not. [7]

Obviously AI pricing — historical, not a live quote

The original pricing page remains reachable and shows a blend of software subscriptions, human-led services and enterprise packages. It does not supply reliably current checkout prices, and it cannot be treated as a live purchasing offer after the 2026 wind-down. [14]

Pricing checked: 30 October 2026. These are legacy web-page displays, not confirmed product availability. An earlier 2021 TechCrunch article quoted plans starting at $75/month; that amount must not be represented as a 2026 price. [9]

How did the business make money?

Confirmed historical monetization included paid no-code software tiers, enterprise sales and managed data-scientist/AI-strategist services. Specific 2026 subscription revenue, agent usage fees, gross margin, net revenue retention or annual recurring revenue were not publicly disclosed. The company was therefore best characterized as software plus human data science, later transitioning to an enterprise AI-agent platform.

Funding, investors and ownership

Financing is one area where the date matters. The company announced $3.6 million in May 2021, and TechCrunch reported an expanded $4.7 million cumulative seed in July 2021. A November 2023 SEC Form D supports another financing event, and O'Shaughnessy Ventures announced its investment in December 2023. The later Zams website stated $10 million raised; this is company-reported, not independently audited. [10][9][11][12][5]

Notable backers: TMV, UTEC, B Capital Group, Golden Gate Ventures, Arka Venture Labs, O'Shaughnessy Ventures; older company material also lists Sequoia Scouts and other investment groups. Accelerator: UC Berkeley SkyDeck, documented by TechCrunch. Company ownership percentages, preferred share rights, comprehensive cap table, post-money valuation and founder control are not publicly disclosed. The 2023 private filing is not itself proof of a current valuation. [9][7]

Valuation, revenue and workforce

Current valuation: not publicly disclosed. Annual revenue, ARR, profitability, margin and audited income: not publicly disclosed. Current employee count after the wind-down: not independently verifiable. Company-directory employee or revenue numbers should be marked as third-party estimates, not official financial disclosures.

Customers and historical market traction

Obviously AI's 2021 press release claimed 3,000+ users and 82,000+ models hosted as of May 2021. These are dated, Obviously AI-reported; not independently verified. They should not be described as active paying customers in 2026. The legacy homepage features named endorsements; those are evidence of attributed testimonials, not independently audited ROI. [10][7]

A company logo shown under “backed by,” “built by,” “integrations,” or “in the news” is not proof of a customer contract. This table intentionally limits examples to named customer testimonies and explicitly described cases. [7][9][18]

Case-study evidence: lending model

The company's anonymized credit-risk case study reports 100,000 data rows, five days to value, 79.2% prediction accuracy and a 7% decline in defaults. A narrative passage on the same page describes identification of 83% of new likely defaulters, which is a different, insufficiently defined metric. These results are Obviously AI-reported; not independently verified and should not be generalized to other datasets. [18]

Industries and target users

Documented or explicitly marketed sectors include marketing, software/SaaS, direct-to-consumer commerce, fintech, insurance, printing, agriculture/yield and consumer gaming. Its original buyer was often a business analyst, marketer or operating manager with usable structured data but without a machine-learning engineering team. Data expertise still mattered for labeling outcomes, assessing leakage, judging fairness and interpreting model results. [9]

Privacy, customer data and model ownership

The April 26, 2026 Zams privacy policy states that Zams did not use customer data to train its own or third-party AI models without explicit written authorization. It also says customer data was processed to deliver customer-defined tasks, with stated deletion-request procedures. This is a statement of documented policy, not an independent audit of actual processing, and predates the wind-down. [19]

The older Obviously AI privacy policy was dated February 4, 2025 and was primarily a general website/service privacy notice. It does not by itself establish a comprehensive contractual right to ML-model export or permanent ownership of every trained artifact. Specific contract and DPA terms would be necessary. [20]

Security and responsible AI

Legacy Obviously AI security pages state SOC 2 Type II coverage, encrypted connections (TLS), encryption at rest (AES-256), Google Cloud hosting, GDPR practices and HIPAA/BAA support. The company-reported compliance assertions were posted before the wind-down; independent current certificate validity and applicability to any successor service were not reverified. Data centers holding third-party certifications are not equivalent to the product itself holding the same certification. [21]

General limitations of AutoML

• Biased, missing or unrepresentative data produce biased predictions.
• Label leakage can inflate test accuracy.
• Correlations are not causal explanations.
• Small sample sizes can make scores unstable.
• Shifting user behavior causes model drift.
• Incorrect targets produce misleading optimization.
• High aggregate accuracy can hide poor subgroup performance.
• Business context and ongoing human oversight remain essential.

These are general ML cautions, not evidence of a documented defect uniquely affecting Obviously AI.

Reviews and user sentiment

The G2 product page showed a small historical review sample (four reviews in the results inspected), insufficient for robust market-wide conclusions. Reviewers praised accessible model creation and helpful hands-on data-scientist support; some said configuration knowledge and limited customization could be challenging. These are user-reported experiences, not product-wide measurements. [22]

Competitors and alternatives

Competitors depend on the period: the original Obviously AI competed primarily in no-code predictive modeling, while Zams competed more broadly with workflow and agentic automation. Following its announced wind-down, use this comparison to identify current alternatives, not as a claim that Obviously AI is actively taking new customers.

The comparison is high-level, based on official platform positioning; specific plan access varies. See vendor sources for Akkio, Pecan, DataRobot and Dataiku. [23][24][25][26]

Obviously AI vs Akkio

Both historically served non-coders with predictive modeling. Akkio now also emphasizes AI-native marketing workflows, connected data, natural-language analysis and forecast/predict functions. For a business purchasing software in 2026, Akkio is a live-market alternative to investigate rather than a simple feature-for-feature successor. [23]

Obviously AI vs Pecan AI

Pecan similarly emphasizes churn, conversions, lifetime value, demand and revenue predictions without a dedicated data-science team, and describes automated data preparation, modeling, validation and deployment. Compared with the historic Obviously AI self-service UI, Pecan's 2026 positioning is closer to an outcome-focused predictive agent. [24]

Obviously AI vs DataRobot

DataRobot positions its offering around enterprise-scale model development, tuning, explainability, operational monitoring, deployment and governance. Its broader MLOps depth may suit mature data teams, while Obviously AI originally prioritized rapid self-service access for comparatively nontechnical buyers. [25]

Obviously AI vs Dataiku

Dataiku combines visual/no-code and full-code ML with collaborative preparation, AI agents and enterprise governance. This is a wider platform category, with greater implementation scope than the original Obviously AI model-building experience. [26]

Obviously AI vs Power BI / Tableau

BI dashboards answer “What happened?” using metrics and visualization. Predictive ML answers “What is likely to happen?” based on a trained model. BI products do support forecasting and extensions, so the boundary is not absolute; they are not exact one-to-one replacements for supervised AutoML in every scenario.

Strengths, limitations and who it suited

Historically best for: business analysts, marketing and sales teams, operations staff, founders and SMBs with clean-enough tabular data and well-defined labels. Historically less suitable for: teams requiring custom deep-learning architectures, absolute control over every ML pipeline stage, or heavily regulated production without additional safeguards. For new buyers in late 2026: assess currently operating alternatives instead of treating archived Obviously AI pages as an active vendor offer.

Frequently asked questions

Short answers are based on dated evidence; a remaining legacy webpage is not a guarantee of commercial availability.

SEO details for Artificial Intelligence Files

Suggested internal-link topics: Obviously AI Tool Profile, Obviously AI Case Studies, No-Code AI Tools, AutoML Companies, Akkio, Pecan AI, DataRobot and Dataiku. These are topic suggestions, not clickable URLs, because the target pages on Artificial Intelligence Files were not independently checked for existence.

Sources and references

Each entry identifies publisher, title, date and original URL. Access/review date for all sources: 30 October 2026. Undated marketing pages are explicitly identified as such; names and dated statements are not assumed to remain true indefinitely.

1. [1] Zams — “Introducing Zams”Nirman Dave / Zams · 11 Feb 2025 · official rebrand
   zams.com/blog/introducing-zams
2. [2] Nirman Dave — personal announcement of wind-downNirman Dave · August 2026 (month corroborated by contemporaneous coverage) · first-person statement
   LinkedIn founder announcement
3. [3] YesPress — “Obviously AI Put a Data Scientist in a Button”YesPress Editors · Aug 2026 · retrospective secondary reporting
   yespress.io/products/obviously-ai
4. [4] Zams — Terms of UseZams · Updated 26 Apr 2026 · legal entity, Delaware status and location
   zams.com/terms-of-use
5. [5] Zams — About UsZams · undated live page · company-reported $10M and leadership
   zams.com/about-us
6. [6] Hampshire College — Founders and $4.7M seedAmy Diehl · 12 Aug 2021 · primary alumni report and founder photograph
   Hampshire College news
7. [7] Obviously AI — Original websiteObviously AI · legacy site, copyright 2025 · historic features and customer testimonials
   obviously.ai
8. [8] Obviously AI — “How No-Code Machine Learning Algorithms Work”Jack Riewe · 14 May 2020 · company technical explainer
   obviously.ai/post/how-no-code-machine-learning-algorithms-work
9. [9] TechCrunch — “Obviously AI … increases seed round to $4.7M”Catherine Shu · 5 Jul 2021 · independent reporting and founder interview
   TechCrunch article
10. [10] Obviously AI — $3.6M seed announcementObviously AI / GlobeNewswire · 3 May 2021 · company press release
    GlobeNewswire press release
11. [11] U.S. SEC — Obviously AI, Inc. Form DSEC EDGAR · 15 Nov 2023 · primary corporate filing
    SEC filing detail
12. [12] O'Shaughnessy Ventures — Investment announcementO'Shaughnessy Ventures / PR Newswire · 15 Dec 2023 · investor announcement
    PR Newswire release
13. [13] Zams — Blog indexZams · latest product-specific dated post identified: 23 Jun 2026
    zams.com/blog
14. [14] Obviously AI — PricingObviously AI · undated legacy page checked 30 Oct 2026 · historical offer limits
    obviously.ai/pricing
15. [15] Obviously AI — Data RequisitesObviously AI · 12 Apr 2021 · structured data requirements
    obviously.ai/data-requisites
16. [16] Obviously AI — AlgorithmsObviously AI · updated 9 Aug 2023 · Edge Sharp AutoML method
    obviously.ai/algorithms
17. [17] Obviously AI — “How We're Democratizing Data Science for SMBs”Obviously AI · historical founder discussion · model validation
    Founder discussion
18. [18] Obviously AI — “Predict Default Rate”Obviously AI · undated official case study · anonymized, unverified outcome metrics
    obviously.ai/case-studies/predict-defaults
19. [19] Zams — Privacy PolicyZams · updated 26 Apr 2026 · customer-data training statement
    zams.com/privacy-policy
20. [20] Obviously AI — Privacy PolicyObviously AI · updated 4 Feb 2025 · legacy data handling disclosures
    obviously.ai/privacy-policy
21. [21] Obviously AI — SecurityObviously AI · official security claims, page carries 2022 and 2024 update labels
    obviously.ai/security
22. [22] G2 — Obviously AI historical product reviewsG2 · reviews dated 2022–2023 · small reviewer sample
    G2 user reviews
23. [23] Akkio — Predictive AI productAkkio · official product page, live 2026 · comparison source
    akkio.com/predict
24. [24] Pecan AI — Predictive AI AgentPecan AI · official product page, live 2026 · comparison source
    pecan.ai
25. [25] DataRobot — Predictive AIDataRobot · official product page, live 2026 · comparison source
    DataRobot predictive AI
26. [26] Dataiku — Product platformDataiku · official product page, live 2026 · comparison source
    Dataiku product platform
27. [27] Fundraising Fox — SEC Form D derived financing summaryFundraising Fox · researched Aug 2026 · third-party filing-derived reading; cross-check SEC
    fundraisingfox.com/companies/obviously-ai

Research notes and disclosure

Editorial standard: “Verified” means supported by a cited primary document or independently corroborated reporting, not that every historical marketing claim was independently tested. Public vendor sites remain accessible after businesses close and may retain old offers; always follow dates and claim classifications. Information reflects the sources reviewed, not investment or purchase advice.