General Assembly's Build AI Agents course was a 32-hour, beginner-oriented program for building no-code and low-code agents. This review checks its former $2,950 price, curriculum, refund terms, credential, student evidence, and—most importantly—whether the course is still available.
Quick answer#
Do not try to buy this course as a consumer today. General Assembly says it stopped new consumer enrollments on April 28, 2026 and is completing an orderly teach-out.[2] The Build AI Agents page remains online but says the course is unavailable and gives no future dates.[1] Historically, the offer cost $2,950 for 32 hours and combined live instruction with labs, integrations, governance, and a final prototype.[1][3]
Verified: September 27, 2026. Best for: nobody as a current consumer purchase, because enrollment has ended. The historical curriculum would have suited nontechnical professionals who valued live structure. Avoid if: you need an enrollable course, named instructor, technical engineering depth, accredited credential, or course-specific graduate evidence.
Is Build AI Agents still available?#
The specific page says, “Currently, this course is unavailable,” and invites visitors to hear when another instance is scheduled.[1] That message is less decisive than GA's regulatory notice, which says the company ceased new consumer enrollments on April 28, 2026 and is concluding consumer operations through a teach-out.[2]
The original URL has not moved to a replacement course. GA's broader catalog still exposes consumer-course marketing, but a retained page is not evidence that enrollment will reopen. Course Report independently lists zero active GA courses while noting that enterprise training continues.[4]
The responsible verdict is therefore historical offer, not current buying option. A waitlist form should not be interpreted as a promised new cohort.
Historical price, duration, and format#
GA published a total price of $2,950, composed in its February 2026 U.S. catalog of a $100 registration fee and $2,850 tuition.[1][3] The scheduled course contained 32 hours across 16 sessions—16 lecture hours and 16 lab hours—normally delivered over eight weeks, with an accelerated one-week format also authorized in the catalog.[3]
That worked out to about $92.19 per scheduled hour. GA also marketed a four-course bundle for $5,900, but that pathway is not currently purchasable through ordinary consumer enrollment.[1]
The course page described live, online, cohort-based delivery. A catalog card also said “On Campus or Online,” leaving the actual format dependent on a scheduled cohort that no longer exists.[1][3]
What the course taught#
The curriculum was organized around four practical areas:
- Agent foundations: the perception-reasoning-action loop, memory, tools, and the no-code/low-code ecosystem.
- Design and building: prompts, intent, context, Flowise, GPT Builder, actions, memory, and multi-agent coordination.
- Integration and scale: Google Sheets, Slack, Zapier, workflow automation, UX, privacy, governance, and access control.
- Application and risk: hallucinations, unauthorized actions, guardrails, testing, monitoring, iteration, and a final agent prototype.[1][3]
The 50/50 lecture-lab split was a genuine strength. It offered more structure than watching disconnected tutorials and put governance and failure risks inside an introductory syllabus.
But the scope remained no-code and low-code. It did not promise depth in Python agent frameworks, custom services, automated evaluation architecture, infrastructure, authentication, secrets, incident recovery, or production software engineering. “Autonomous agents” was an ambitious label for a 32-hour beginner course.
Prerequisites and software costs#
GA called the course beginner-friendly, required no technical background or prework, and targeted business, operations, product, innovation, consulting, and entrepreneurial learners.[1]
Published setup requirements included a laptop with administrator access, at least 8 GB RAM and 40 GB free storage, stable internet, and a webcam. GA said learners could use free tiers from OpenAI, Claude, Perplexity, Gemini, or Copilot and that premium subscriptions were not required.[1]
That claim did not establish that every integration and final project would fit indefinitely inside free quotas. Hardware, model/API usage, Zapier or other automation tiers, hosting, and post-course operation remained potential extra costs. For an active agent that needs deployment and uptime rather than a class, compare the managed FlyHermes path. For open-source control, start with the Hermes Agent installation guide.
Refund and cancellation terms#
There was no simple universal money-back guarantee. GA's FAQ and catalog say the enrollment agreement and buyer's jurisdiction control the refund schedule.[3][5]
Under the general U.S. catalog terms, a student could cancel before the first class without tuition penalty, normally less the $100 registration fee, by sending written notice. Later refunds depended on state-specific completion bands; some states allowed prorated refunds only through 40%, 50%, 60%, or 75% completion.[3]
If GA postponed or cancelled a course outside specified force-majeure situations, the catalog offered a later date or refund of paid course fees.[3] Buyers should have read the actual enrollment agreement rather than relying on a sales-page summary. These historical rules do not create a right to enroll now.
Who was behind the course?#
General Assembly Space, Inc. was founded in 2011, joined The Adecco Group in 2018, and operates within Adecco's LHH portfolio.[6] That establishes a real company and substantial training history.
The problem at course level was instructor transparency: the unavailable page did not name the person who would teach Build AI Agents. It offered only generic assurances about industry practitioners.[1] Since live instruction was a major reason to pay a premium, the inability to evaluate a cohort's instructor before purchase was material.
The local Google/YouTube transparency scan found one General Assembly Space Inc. creative, ID CR10670581273199640577, first observed March 16 and last observed August 26, 2026.[7] The archive did not capture copy or destination, so it cannot establish that the creative promoted this course. Its later date could reflect enterprise training or another GA offer—not continued consumer enrollment.
Credential value#
Successful students were promised a LinkedIn-verified digital badge.[1] GA also states that it is not accredited by an accreditor recognized by the U.S. Department of Education, and its programs are not eligible for federal student aid.[5]
A verified badge can confirm completion under the provider's rules. It is not a degree, professional license, academic credit, or proof of production engineering ability. The prototype and documented workflow would have mattered more than the badge alone.
Complaints and independent evidence#
There is very little independent evidence for this exact course. No credible body of course-specific Reddit discussions, public capstones, named alumni case studies, or verified graduate reviews was found. The testimonial displayed on the course page came from a Digital Marketing Short Course graduate, not a Build AI Agents student.[1]
Broader GA reviews are mixed but cannot validate this new syllabus. Course Report shows a historical 4.31/5 rating from 714 reviews, while Trustpilot displayed praise for practical instructors alongside complaints about instructor continuity and price.[4][8] Those are signals about the provider as a whole, not measured outcomes for Build AI Agents.
An editorial review site gave the course 4.4/5 but did not expose a named reviewer, verified student sample, cohort date, or transparent methodology.[9] Treat that as a syllabus summary, not alumni proof.
Better alternatives now#
Because this course cannot be purchased, the useful comparison is not whether its former curriculum looked attractive. It is what outcome you need now:
- To learn by shipping, follow Learn AI With an Agent and build one real workflow.
- For technical control, use Hermes Agent, persistent memory, and browser automation.
- For structured technical instruction, compare the Agentic AI Engineering Bootcamp review.
- For university-branded, engineering-oriented material, read the Stanford Agentic AI Program review.
- If the blocker is maintenance rather than education, use a managed deployment instead of buying a discontinued course.
Verdict#
Build AI Agents had a coherent beginner syllabus, a useful lecture-lab split, and practical integration topics. Its weaknesses were high hourly cost, no named instructor, limited engineering depth, and almost no course-specific independent evidence.
Those are now secondary. General Assembly has ended new consumer enrollment. Keep this page as a factual historical review, not a recommendation to join a waitlist or hunt for an unofficial checkout. Reconsider only if GA publishes a new legal offer with dates, instructor identity, current curriculum, and an applicable enrollment agreement.
Sources#
- General Assembly — Build AI Agents
- General Assembly — regulatory information and teach-out notice
- General Assembly — February 2026 U.S. catalog
- Course Report — General Assembly
- General Assembly — FAQ
- General Assembly — about
- Google Ads Transparency — General Assembly creative
- Trustpilot — General Assembly
- AI Course Review — Build AI Agents