Course 3.1 · Curriculum 2026.1
The three places AI can run, and how to choose
Separate a conversational agent, a model called from Apex, and a model in a pipeline. Most teams only know the middle one and pick it for the wrong problems.
- Estimated effort
- 3 hr
- Lessons
- 27
- Modules
- 5
- Material
- 3 hr
Estimated effort is the sum of the lesson and lab times in the course outline, for this version of the course.
This course is planned and not on sale. The curriculum below is the plan, and it may change before the course opens.
See the courses on sale01What you will learn
6 capabilities, stated as things you can do
Each one is what you should be able to do at the end, not what the course covers. If an outcome is not testable, it is not an outcome.
- Name the three tiers, and place any AI request into one of them in a five-minute conversation.
- Say what each tier costs to build, to run and to change.
- Explain why a conversational agent is the wrong shape for a nightly enrichment job.
- Explain why an off-platform pipeline is the wrong shape for a user-facing answer.
- Sketch the data path for each tier, including where identity and permissions are enforced.
- Write a one-page recommendation a non-technical sponsor can approve.
02Curriculum
5 modules, 27 lessons
Every lesson shows its length and its type, from the course outline.
5 modules · 27 lessons · 3 hr of material
Salesforce’s own leadership has said customers test agents in the high eighties and watch it fall to fifty or sixty percent in production. This course is about the gap, and about the three shapes an AI feature can take before you write anything.
- VideoThe accuracy that drops from the high eighties to the fiftiesA public number, not an internal one. What it says about demos versus production before you have built anything.Preview6 min
- VideoWhat that gap is actually made of5 min
- VideoThree tiers, one decision5 min
- VideoWhere identity and permissions live in each5 min
- ReadingThe vocabulary this course uses: runtime, named credential, worker5 min
A conversational agent on an external agent runtime. The session carries identity for you, which is exactly why teams reach for it even when the problem is not conversational.
- VideoTier one: the conversational agent6 min
- VideoWhat a conversational agent costs to build, run and change5 min
- VideoWhere identity is enforced: the session inherits it5 min
- ReadingWhen this tier is wrong: the nightly job nobody should chat with5 min
- LabLab: sketch the data path for a conversational agent12 min
A model called from inside the platform through a named credential. The one most teams already know, and the one they reach for when it is the wrong shape.
- VideoTier two: a model called from Apex6 min
- VideoWhat a platform call costs to build, run and change5 min
- VideoWhere identity is enforced: the caller’s own context5 min
- VideoWhy most teams reach for this tier by default5 min
- ReadingWhen this tier is wrong: the answer nobody is waiting for5 min
- LabLab: sketch the data path for a platform call12 min
A model in an off-platform worker pipeline. Nobody is waiting on this one, which is the point, and identity has to be handed in rather than inherited.
- VideoTier three: a model in a pipeline outside the platform6 min
- VideoWhat an off-platform pipeline costs to build, run and change5 min
- VideoWhere identity is enforced: handed in, never inherited5 min
- ReadingWhen this tier is wrong: the answer a customer is waiting for5 min
- LabLab: sketch the data path for an off-platform pipeline12 min
A decision path you can run in a meeting, three worked examples, and the one-page recommendation a non-technical sponsor can actually approve.
- VideoThe decision path, in five questions5 min
- VideoWorked example: a support answer a customer reads5 min
- VideoWorked example: a nightly enrichment over 250 records5 min
- VideoWorked example: a field technician asking a question by voice5 min
- LabLab: write a one-page recommendation for your own case18 min
- QuizCheck: choosing a tier12 min
04Before you start
What you need first
Assumed knowledge and setup
- No AI experience assumed. This is the entry point to the track.
- Salesforce familiarity at the level of course 1.1.
- No org required. This course is conceptual and is the shortest in the catalogue.
Before this course
No other course is needed first.
Courses often taken next
- 3.2 · Buy versus build: when the answer is a lookup and AI would only add cost
- 3.3 · Prompt Builder and Agent Builder in practice
- 3.4 · Calling a model from Apex safely
Each is a separate course, bought on its own.
05Who teaches it
Turni Saha
AI Adoption Lead, PMP
ShareCo Academy, and AI adoption for teams that are not technical.
Project Management Professional, issued 2024
06Reviews
No learner reviews yet
This course has not been taken by enough people to publish an honest rating, and we will not print invented quotes on a page that sells verification discipline. Here is what we can evidence instead.
- Assessment
- Lesson checks onlyThis course carries lesson checks and issues no certificate.
- Curriculum version
- 2026.1You enrol into a version. Content changes do not move you mid-course, and your required-lesson count is snapshotted at enrollment.
- Refunds
- 14 daysAn individual purchase can be refunded within 14 days if you have completed less than 20% of the lessons. Team seats are refundable before they are assigned. From the Academy terms.
Start 3.1
This course is in development and not yet open for enrolment. The courses open now are listed in the catalogue.
- 3 hr estimated effort, labs included
- A ShareCo certificate is not a Salesforce certification