Short answer: Choose an AI SDR by testing it against your own stack: what it writes to your CRM, whether its research carries sources and dates, which signals it publishes, and what it checks before a mailbox sends. Outlevo, our product, writes only to fields it owns in HubSpot and needs at least three dated observations per account card [[1]](#sources) [[2]](#sources).
Who this is for
This guide is for a B2B SaaS team that already runs a CRM and has account executives (AEs) who take over once a prospect says yes. You want software to research accounts, send outreach and pass interested replies to a rep. You also want to know what to ask before a trial starts.
It helps to price the alternatives first. One published 2026 estimate puts the fully loaded annual cost of one in-house SDR in the US at $115,000-$165,000, covering salary, benefits, tech stack, onboarding and ramp losses [[3]](#sources). One October 2026 guide, quoting Clutch, puts the average cost per qualified B2B appointment in 2025 at $550-$1,700 [[4]](#sources). Both are third-party figures, not forecasts for your team.
For a side-by-side of the three models, read SDR vs sales engagement platform vs outbound agency. For a definition of the term, see AI SDR. If you are earlier than that, with few customers and no settled process, start with early stage B2B outbound.
Outlevo is our product. We name it here as one option among several. Where we describe other vendors, we use only what they publish on their own pages, with the date we checked.
Steps
Write the handoff first
Before any demo, write down what an AE must have in hand when a lead arrives: the company, the owner, why now, the evidence and the next step. Then judge each tool by whether it delivers that into your CRM, not by how its sample email reads.
For reference, here is how Outlevo handles handoff. It reads CRM owners so hot leads go to the right rep [[1]](#sources). It can answer an interested reply with your Calendly link, with the person's name and email filled in [[5]](#sources). On autopilot the answer goes at once. Otherwise it waits in Replies for you to send [[5]](#sources). Each booking with one of your contacts counts as a meeting booked and is written to your CRM when one is connected [[5]](#sources). Cal.com works the same way [[6]](#sources).
Sort the field into categories
Tools that get called "AI SDR" differ in what they are built around. The groups below come from what each vendor publishes about itself.
- Data layers inside your CRM. Bitscale calls itself the GTM data layer for your CRM, living inside Salesforce and HubSpot, with AI agents, 100+ data sources and live buying signals (checked 2026-11-09) [[7]](#sources).
- Signal-led outreach tools. AiSDR, 11x and Gojiberry each publish a list of signals, covered in a later step [[8]](#sources) [[9]](#sources) [[10]](#sources).
- Email-writing tools and agents. Lavender positions itself as the "Email intelligence company", offering an AI email coach and an AI sales agent called Ora (read 2026-11-09) [[11]](#sources). Ora is $500 per agent per month at launch (read 2026-10-27) [[12]](#sources). Lavender Teams starts at $8,500 (read 2026-10-27) [[13]](#sources).
- CRMs with AI add-ons. Close lists an AI Call Assistant at $50/month per account, plus $0.02 per recorded minute (read 2026-11-16) [[14]](#sources) [[15]](#sources). Your CRM is a separate decision. Outlevo's CRM pricing comparison lists the published plans.
For the full list of published AI SDR tools with pricing, see ai sdr tools. If you are weighing a sequencing tool rather than an agent, see the Reply.io alternative page. Enrichment vendors are compared on data and enrichment tools.
Check the CRM connection
Most of the work of an AI SDR ends up in your CRM, so test that connection hardest. The question is not "does it integrate". It is what it reads, where it writes, and who approves the mapping.
With HubSpot, Outlevo reads your companies, deals and owners [[1]](#sources). It writes its status, fit score and next step to each company in fields it owns [[1]](#sources). Specifically:
- Your standard HubSpot fields are never edited [[1]](#sources).
- Nothing is written until an owner or admin approves the field mapping [[1]](#sources).
- It can write a note, a task and the booked meeting on the company, each one switchable in the mapping [[1]](#sources).
- Contact import is optional. It stays off until an admin turns it on with a stated purpose [[1]](#sources).
- Connecting and syncing a CRM uses no credits [[1]](#sources).
The details are on the HubSpot integration page.
Salesforce is different, and the page says so plainly. Outlevo reads Salesforce accounts, opportunities and users, and writes only its own fields, Readiness Cards and tasks, through the Outlevo package [[16]](#sources). The connection is a pilot. It needs the package in your org, which Outlevo installs with you, and there is no AppExchange listing [[16]](#sources). Salesforce sends no webhooks to Outlevo, so a rep's Accept or Reject on the Account is read by the 15-minute sync [[16]](#sources). See the Salesforce integration page.
If you use another CRM, Outlevo's published routes are a CSV import of up to 200,000 rows [[17]](#sources), the API, and Zapier, Make or n8n workflows that use the API [[18]](#sources) [[19]](#sources) [[20]](#sources). The REST API has 249 paths across resources that include companies, contacts, campaigns, handoffs and webhooks [[21]](#sources). Everything in the app is available in it [[22]](#sources).
Whatever tool you test, ask these questions:
- Does it write into your standard fields or into fields of its own?
- Who approves the field mapping, and is anything written before that?
- What happens in the tool when a rep rejects a lead?
- How long does a rep's decision take to travel back?
Check how it learns your ideal customer
An AI SDR that targets the wrong accounts only fails faster. Ask where its idea of your ideal customer comes from.
In HubSpot, Outlevo learns your ideal customer from the deals you actually won and lost, then keeps each company record up to date with what it found [[1]](#sources). A release on 2026-10-19 added the option to import past deals from CSV or your CRM and get a win/loss analysis that proposes ICP hypotheses [[23]](#sources). The analysis counts companies, not deals, and labels small samples: under 10 won companies is very low confidence [[24]](#sources). AI-proposed segments are labelled experimental until a person approves them [[24]](#sources).
Outlevo also asks which countries you sell into and never proposes markets [[24]](#sources). Look for the same discipline elsewhere. A tool that picks your markets for you is making a decision that belongs to you.
Check how it researches accounts
Research is where an AI SDR can quietly make things up. Ask to see the evidence behind a claim in a draft.
Each qualified account in Outlevo gets a Readiness Card with fit, intent and data quality scores [[2]](#sources). A card needs at least three observations, each with a source and a date [[2]](#sources). Facts read from a website are kept only when their quote is found on the page [[2]](#sources). Guesses are labelled as hypotheses and never presented as findings [[2]](#sources).
In credits, qualification costs 3, light qualification 1, contact discovery 4, deep research 30 and email verification 1 [[25]](#sources).
Compare the signals
Many teams act on changes such as a hire, a funding round or a visit to the pricing page. Signal lists are easy to publish and hard to compare, so check them against the triggers you act on. This is what four vendors list:
- AiSDR lists job changes, new hires and funding, with LinkedIn signals and website visitors on team plans (checked 2026-10-22) [[8]](#sources).
- 11x lists job changes, funding rounds, executive hires, tech stack shifts, website visitors and G2 buying signals (checked 2026-10-22) [[9]](#sources).
- Gojiberry claims 15+ buying and social signals, such as engagement with competitors' LinkedIn content, job changes, new hires, funding, hiring spikes and website visitors (checked 2026-10-22) [[10]](#sources).
- Bitscale lists AI agents, 100+ data sources and live buying signals (checked 2026-11-09) [[7]](#sources).
A list on a pricing page shows what a vendor says it covers. It does not show how often a signal fires, or how accurate it is. Ask for a sample from your own target list.
None of these lists names product usage. If usage data from your own product matters to you, ask each vendor directly how it would come in.
Outlevo's website-visitor signal works at company level. One script on your site shows which companies visit, which pages they read and whether they fit your ICP [[26]](#sources). It identifies the organization behind a visitor's network, never a person, and sets no cookie [[26]](#sources). It costs 1 credit per company identified per day, however many pages it reads [[26]](#sources).
Check the sending side
Good research does nothing if the mail lands in spam. Ask what the tool verifies before it sends, and who owns the mailboxes.
Outlevo writes and schedules the emails, and Smartlead sends them from the mailboxes in your Smartlead account [[27]](#sources). Before a campaign sends, every mailbox must be connected and warmed up, and its domain must have SPF, DKIM and DMARC checked in the last 7 days [[27]](#sources). Sent, reply, bounce and unsubscribe events come back by webhook [[27]](#sources).
If you would rather not touch your main domain, Outlevo's managed mailboxes buy sending domains, create the mailboxes through Zapmail, set up DNS and warm them up [[28]](#sources). A new mailbox sends nothing until it is warm: at least 14 days, usually about 3 weeks [[28]](#sources).
On LinkedIn, each person connects their own account [[29]](#sources). Automatic sending comes with Self-Serve and Managed, or with a LinkedIn seat on Builder. Otherwise the invitations and messages are drafted for you to send by hand [[29]](#sources).
Mailbox providers set their own bars. Google says "email senders who send more than 5,000 messages per day to Gmail accounts must meet the requirements in this section. Set up SPF and DKIM email authentication for your domain." (checked 2026-10-22) [[30]](#sources). Microsoft says "For domains sending over 5,000 emails per day, Outlook will soon require compliance with SPF, DKIM, DMARC." (checked 2026-10-22) [[31]](#sources).
On unsubscribes, RFC 8058 "describes a method for signaling a one-click function for the List-Unsubscribe email header field." (checked 2026-10-22) [[32]](#sources). Ask any vendor how it handles unsubscribes and whether it supports that header.
For a sequence length to plan around, see how many follow-ups a cold email sequence needs.
Check the rules and the controls
An AI SDR can send in your name at a volume no person would. Two things matter here: the law that applies to your recipients, and the controls the tool gives you.
On the law, the FTC's CAN-SPAM guide says: "The law makes no exception for business-to-business email." (checked 2026-10-22) [[33]](#sources). The UK ICO's guidance on electronic mail marketing says: "You can email or text any corporate body (a company, Scottish partnership, limited liability partnership or government body)." (checked 2026-10-22) [[34]](#sources). GDPR Article 6(1)(f) is one lawful basis for processing personal data. It covers cases where "processing is necessary for the purposes of the legitimate interests pursued by the controller or by a third party" (checked 2026-10-22) [[35]](#sources). It is not a blanket permission to email, and whether it fits your outreach is a question for your counsel. Rules differ by place and by type of recipient. Treat these quotes as starting points.
On controls, Outlevo sends nothing outside a policy that a person approved [[36]](#sources). Policies set who can be contacted, on which channels, how often and with what budget [[36]](#sources). Every external action is checked against the approved policy before it runs, and recorded in a ledger [[36]](#sources). Kill switches pause a campaign, a channel or the whole workspace at once [[36]](#sources).
On data, every customer table is protected by row-level security in the database, and personal data such as email addresses is encrypted at the field level [[37]](#sources). More is on the security page.
If your team builds on top of the tool, test keys run against a sandbox with fake companies and contacts and are not charged [[22]](#sources). Metered API calls quote their credit cost first and run only when confirmed [[22]](#sources). Zapier, Make and n8n connect through the API and signed webhooks, at 1 credit per delivered event [[18]](#sources) [[19]](#sources) [[20]](#sources). There is no Outlevo app in Zapier or Make: Zapier uses Webhooks by Zapier, and Make uses a Custom webhook and an HTTP module [[18]](#sources) [[19]](#sources). Outlevo's developer hub has the details.
Price the whole workflow
Compare total cost, not the headline plan. Add the plan, the usage charges, the sending infrastructure and the CRM seats you already pay for.
Outlevo's plans are:
- Free, at USD 0 per month [[25]](#sources).
- Builder, USD 29 per month with 320 credits a month [[25]](#sources).
- Growth, USD 149 per month with 3000 credits a month [[25]](#sources).
- Self-Serve, USD 199 per month with 1600 credits a month [[25]](#sources).
- Managed, from USD 999 per month on a contract [[25]](#sources).
CRM sync and sending use no credits [[1]](#sources) [[27]](#sources). Credits go to actions such as research and qualification, listed above. Self-serve plans and credit top-ups go through Stripe Checkout, with sales tax [[38]](#sources). See the pricing page for the current table.
For a fair comparison, ask every vendor for the same four numbers: monthly platform cost, usage charges for your volume, sending cost, and the time your team spends reviewing output.
What to check
Use this list when you score tools side by side. If a vendor cannot answer a row in writing, treat that row as unanswered.
| Area | Ask the vendor | What a clear answer looks like |
|---|---|---|
| CRM write-back | Which fields does it write to, and which does it never touch? | A named list of fields, with your standard fields left alone |
| Approval | Who approves the mapping, and when does writing start? | A named role, and nothing written before approval |
| Handoff | How does a hot lead reach the owning rep? | Owner taken from the CRM, with a task or note the rep can see |
| Rep feedback | What happens when a rep rejects a lead? | A defined path back into the tool, with a stated delay |
| ICP | Where does the target profile come from? | Your won and lost deals, with sample size stated |
| Evidence | Can I see the source and date for each claim in a draft? | A source and a date next to each fact, guesses labelled as guesses |
| Signals | Which triggers fire on my target list, and how often? | A sample from your own accounts, not a feature list |
| Sending | What is checked before a mailbox sends? | Authentication, warm-up and domain checks, stated |
| Unsubscribes | How are opt-outs handled across channels? | A documented path that stops further messages |
| Controls | Can I pause one campaign, one channel or everything? | Separate switches at each level |
| Audit | Is every outbound action logged? | A record you can read and export |
| Cost | What do I pay at my volume, all in? | A worked example from your numbers |
Run each shortlisted tool on the same twenty accounts from your own CRM. Have an AE read the output and say which leads they would call.
Next step
Start with your own data. Outlevo's free ICP assessment reads up to 5 public pages of your website [[39]](#sources). It returns facts labelled as verified observations or AI inferences, and up to 5 experimental segment hypotheses [[39]](#sources). It also asks which countries you sell into [[40]](#sources).
If you use HubSpot, the next step after that is the HubSpot integration. An owner or admin connects it and approves the field mapping [[1]](#sources). Compare what you see with the other tools on your shortlist, using the table above.
Sources
- HubSpot integration
- Readiness Cards
- Stealth Agents, cost of hiring an SDR (2026)
- Salesar, appointment setting costs (October 2026), citing Clutch
- Calendly integration
- Cal.com integration
- bitscale.ai
- AiSDR pricing
- 11x signals and triggers
- gojiberry.ai/llm
- Lavender: positioning
- Lavender: pricing
- Lavender: pricing
- Close: pricing
- Close: pricing
- Salesforce integration
- CSV and Excel integration
- Zapier integration
- Make integration
- n8n integration
- Outlevo API reference
- Developer hub
- Release 2026-10-19
- ICP from evidence
- Outlevo pricing
- Website visitors integration
- Smartlead integration
- Managed mailboxes integration
- LinkedIn integration
- Google: Email sender guidelines
- Microsoft: Outlook requirements for high-volume senders
- RFC 8058: Signaling One-Click Functionality for List Email Headers
- FTC: CAN-SPAM Act compliance guide for business
- ICO: Electronic mail marketing (PECR)
- GDPR Article 6(1)(f): lawfulness of processing
- How Outlevo works
- Security
- Release 2026-10-19
- Free ICP assessment
- Release 2026-10-19